CN108564109A - A kind of Remote Sensing Target detection method based on deep learning - Google Patents

A kind of Remote Sensing Target detection method based on deep learning Download PDF

Info

Publication number
CN108564109A
CN108564109A CN201810235045.5A CN201810235045A CN108564109A CN 108564109 A CN108564109 A CN 108564109A CN 201810235045 A CN201810235045 A CN 201810235045A CN 108564109 A CN108564109 A CN 108564109A
Authority
CN
China
Prior art keywords
network
model
remote sensing
layer
training
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201810235045.5A
Other languages
Chinese (zh)
Other versions
CN108564109B (en
Inventor
侯春萍
夏晗
杨阳
管岱
莫晓蕾
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tianjin University
Original Assignee
Tianjin University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Tianjin University filed Critical Tianjin University
Priority to CN201810235045.5A priority Critical patent/CN108564109B/en
Publication of CN108564109A publication Critical patent/CN108564109A/en
Application granted granted Critical
Publication of CN108564109B publication Critical patent/CN108564109B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Computational Linguistics (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

The Remote Sensing Target detection method based on deep learning that the present invention relates to a kind of, including：Associated data set is built using remote sensing images：After remote sensing images are classified and are marked, image data set and the class label generated by markers work；It builds based on the panchromatic sharpening model for generating confrontation network；The target detection model based on depth convolutional neural networks is built, end-to-end training is carried out to model by the methods of backpropagation and stochastic gradient descent；End-to-end test is carried out to the model built.The present invention has the advantages that accuracy is high.

Description

A kind of Remote Sensing Target detection method based on deep learning
Technical field
The present invention relates to the fields such as remote sensing image processing, deep learning, pattern-recognition, and depth is based on more particularly, to one kind Study, the method that panchromatic Edge contrast and target detection are carried out to spectrum picture using generation confrontation network.
Background technology
Due to the limitation that signal transmission wave band and imaging sensor store, most of remote sensing satellites, which only provide, has EO-1 hyperion Multispectral (MSI) image of resolution ratio and panchromatic (PAN) image of high spatial resolution.Advantage using two kinds of images is mutual It mends, is fused into the amalgamation remote sensing image with clear spatial detail and abundant spectral information, this integration technology, also referred to as Panchromatic sharpening technique.
Currently, the important Shift Method of panchromatic sharpening method of remote sensing fields mainstream, multiscale analysis method etc..Component Shift Method It is right on color space domain mainly by principal component analysis, Schimidt orthogonalization and the methods of intensity, tone, saturation degree transformation Spectrum picture is converted, and the spatial information channel of multispectral image is replaced with full-colour image, blending image is obtained by inverse transformation.
Multiscale analysis method refers to based on the approach such as wavelet transformation, laplacian pyramid and multi-scale geometric analysis, Source images are decomposed into a sequence decomposition coefficient with multiresolution analysis tool, then merge these decomposition coefficients by certain Criterion is merged into the decomposition coefficient of blending image, and blending image is obtained finally by multiresolution analysis tool inverse transformation.
In recent years, the development of the appearance based on large-scale data and deep neural network, deep learning method become machine The important research direction of device learning areas.Thought based on depth learning technology and game theory can be generated high by low-dimensional feature The generation confrontation network (Generative Adversarial Networks, GAN) of dimension sample can be introduced into for panchromatic sharp Change process, GAN is by generation network model and differentiates that network model is constituted.Generating model can help to generate correlated samples number According to, and discrimination model may determine that the validity of sample, the two training simultaneously, generating model then constantly reinforces, by constantly changing In generation, makes generation sample become closer to authentic specimen.
As pattern-recognition remote sensing fields an important application, based on the multiple types multiscale target under remote sensing images How detection and a key technology that identification is the fields such as geophysical surveying, military surveillance and accurate strike, improve target detection Precision, also the always research hotspot and difficult point in remote sensing application field, there is important military and civilian value.With high score The fast development of resolution remote sensing technology is able to build extensive high-definition remote sensing image data collection, more intelligent to develop Remote sensing image object detection system provides possibility, and target effective feature is extracted from mass data just becomes remote sensing images application Key technology.
Traditional detection algorithm is almost to complete classification and detection work in certain given feature base.Extraction Feature and detection model play the role of model vital as the two big key factors for determining detection result.This will Ask has strict requirements to the feature of input, and finds the detection model of matching this feature.
However above-mentioned requirements are undoubtedly complicated and take, and it is strongly depend on the spy of professional knowledge and data itself Sign fully is excavated between large-scale data additionally, it is difficult to learn an effective disaggregated model from large-scale data It is interrelated.
With the development of deep learning method so that initial data is realized end-to-end (End-to-End) as input Learning process is possibly realized.Deep layer artificial neural network has very strong feature learning ability, deep learning model learning to obtain Characteristic has former data more essential representativeness, passes through the trained target based on depth learning technology of large-scale data Detection model can more extract its abundant internal information, be conducive to visualization and classification problem processing.Therefore it is based on convolutional Neural Network can design a kind of method for capableing of automatic learning characteristic, and by the study to mass data itself, acquisition wherein most has The further feature of effect, and by establishing relative complex network structure, the association between abundant mining data.
Invention content
The Remote Sensing Target detection method based on deep learning that the object of the present invention is to provide a kind of, the present invention are applied to The generation of panchromatic sharpening fights network model, can realize the expansion of information content in remote sensing images；Applied to target detection Depth convolutional neural networks model, accuracy higher, real-time is more preferable, and robustness is stronger.For achieving the above object, this hair It is bright to adopt the following technical scheme that：
A kind of Remote Sensing Target detection method based on deep learning, includes the following steps：
1) remote sensing images are utilized to build associated data set：After remote sensing images are classified and are marked, image data set and By the class label that markers work generates, and training set and test set are divided, is used for subsequent network training and test；
2) it builds based on the panchromatic sharpening model for generating confrontation network：The generation network G of GAN is that can learn from making an uproar at random Image x in data set described in sound vector z and 1, to the mapping of the sample image y generated, i.e. G：{ x, z } → y generates mould Type, which is taken, increases the U-Net structures for redirecting connection, is divided into coding layer and decoding layer two parts, often encodes one layer, characteristic pattern is long Halve with width, the feature number of plies increases half, often decodes one layer, and the length and width of characteristic pattern double, and feature number of plies increase doubles and right The coding layer answered, is concatenated by channel, then carries out deconvolution processing；Based on the convolutional neural networks CNN for classification, design Differentiate that network model, the network are designed to containing there are one concatenation layer and four layers of convolutional layers；
3) the target detection model based on depth convolutional neural networks is built：It is given birth to according to the candidate region of algorithm of target detection At, feature extraction, classification, four steps of position refine, within above-mentioned steps unification to a depth network frame, Concurrent operation in GPU, feature extraction is using residual error network ResNet as basic sorter network, wherein including several convolutional layers and line Property unit R eLU, design section generate network structure all possible candidate frame is sentenced on the characteristic pattern extracted Not, by sharing convolution, the marginal cost for calculating Suggestion box is reduced, by backpropagation and stochastic gradient descent method to model Carry out end-to-end training；
4) end-to-end test is carried out to the model built：Based on the data set built in step 1, training objective detection Model and test model.
Compared with prior art, promotion of the invention and advantage are：
One, different from all existing thinkings of Remote Sensing Target detection method, the present invention innovatively proposes to be based on The cascade method for first carrying out panchromatic Edge contrast and carrying out target detection again of deep learning method.Carry out the process of panchromatic sharpening In, GAN can utilize the higher-dimension further feature implied in depth convolutional neural networks extraction large-scale data, structure also can Farthest reduce the information loss of convolution process.The remote sensing images after panchromatic sharpening are carried out, there is clearly spatial detail With abundant spectral information, there is the high spatial resolution and high spectral resolution relative to full-colour image and spectrum picture Characteristic, the basis that can improve remotely-sensed data in data set utilize the abundance of information, and the promotion of spatial resolution is for Small object Detection more have significant practical applications.Work is directly detected end to end to image by trained model, more To be efficient, time and computing redundancy degree are lower.
Two, different from the conventional method of existing Remote Sensing Target detection, the present invention innovatively proposes to be based on depth The target detection network model of convolutional neural networks.Relative to the manual feature extraction such as HOG features, feature can directly use data The middle characteristic pattern that is obtained after convolutional neural networks indicates, since convolution operation has a translation invariant shape, in characteristic pattern not only The classification information of object is contained, also includes the location information of object, is had so the classification results of feature and position return Better accuracy and stronger universality.The present invention uses Area generation network, region to recommend to be also placed in network to complete, from The overall process that feature extraction detects to the end is all completed in one network, and speed promotes higher, while it is related to solve fitting Problem.
Description of the drawings
Fig. 1 is the flow chart tested needed for the present invention.
Fig. 2 is the generation schematic network structure of panchromatic sharpening.
Fig. 3 is the panchromatic sharpening design sketch of remote sensing images, (a) panchromatic remote sensing image data；(b) spectral remote sensing picture number According to；(c) amalgamation remote sensing image data；(d) amalgamation remote sensing image data of the present invention.
Fig. 4 is Area generation network structure.
Specific implementation mode
To keep technical scheme of the present invention clearer, the specific embodiment of the invention is further described through below. As shown in Figure 1, the present invention implements according to the following steps：
1. the extensive remote sensing image data collection of structure
The present invention selects SpaceNet on AWS, NWPU VHR-10, United States Geological Survey USGS etc. disclosed in network Remote sensing images collection is detected the data set structure of task.
NWPU VHR-10 data sets are a ten publicly available type geographical space object detection data sets.This ten class Article is aircraft, ship, oil storage tank, harbour and bridge etc., including the label of target and its mark in high-definition picture and figure File.
SpaceNet is the extensive remote sensing image data collection for being hosted in Amazon companies AWS cloud service platforms, is DigitalGlobe, CosmiQ Works and NVIDIA are completed jointly, it includes the on-line storage library of satellite image and The satellite image data that the training data marked is the high-resolution published, is exclusively used in training machine learning algorithm is flat Platform.In addition to this, the present invention is also in relation with Chinese Academy of Sciences's geographical spatial data cloud platform, United States Geological Survey (USGS) and Google Required data set is trained and tested to the related remotely-sensed data of company to build.
Image data in above-mentioned data set is pressed 4：1 ratio has been divided into training set and test set.The present invention to its into After row classification and mark, the present invention makes image data set and label according to the format of PASCAL VOC challenge matches, for follow-up Network training and test.
2. building based on the panchromatic sharpening model for generating confrontation network
Based on the related Remote Sensing Database built in 1, the generation confrontation for the panchromatic sharpening of remote sensing images is built and trained Neural network, the step are to provide the remote sensing images number with high spatial resolution and spectral resolution for subsequent detection According to.Generation for panchromatic sharpening fights network, by generation network and differentiates that two networks of network are constituted, generates network and differentiation Network is usually made of the multitiered network comprising convolution and (or) full articulamentum.It is carried out by the network structure to excellent more Secondary experiment test, the present invention build the convolutional neural networks using based on U-Net networks as generation network.
Generation network is built using the U-Net frameworks of full convolutional coding structure, and builds the differentiation network of different size receptive field Framework.Down-sampling is realized by using the convolution kernel in U-Net networks, can not only reduce the redundancy of operation, and The abstract characteristics of target can also be extracted to a certain extent；Convolution operation is carried out with a variety of different convolution collecting images, can be obtained Response on to different IPs, the feature as image.The image array of input passes through after convolution kernel (kernal) convolution algorithm An obtained new image array, i.e. characteristic pattern (feature map).
Latter linked unit can keep characteristic pattern Scale invariant herein, in addition, by the pond layer in network instead of spy Levy the convolutional layer of figure Scale invariant；The full articulamentum in network is deleted, the up-sampling of image is realized with warp lamination, here can It is enough to handle the feature that shallow-layer convolutional layer is exported with deep layer convolutional layer, improve the accuracy of feature extraction.
The full-colour image x in random noise signal z vector sum databases is inputted in generating network as described above, will be given birth to The image data y generated at network is as the input for differentiating network, i.e. G：{ x, z } → y, by training, generating sample cannot be by Differentiate that network model is determined as vacation.And differentiate network model D, by training, can complete to differentiate generation sample as well as possible Classification problem.
The training objective of GAN can be used following loss function formula to indicate, wherein x is the conventional images of input, and y is output Sample image, z be random noise vector：
It generates model and discrimination model all and is using the structure of convolutional layer-batch standardization-linear unit, GAN handles figure The detail sections such as the structural information as medium-high frequency generate model to make the minimized target during training, and differentiate mould Type makes its maximization, i.e.,
G*=argminGmaxDLcGAN(G,D)
In the generating process of final sample image, the full-colour image of input and the blending image of output bottom having the same Structure shares the position of projecting edge.In order to make generation model obtain the information, generation model, which increases, redirects connection, uses The overall structure of U-Net.
U-Net is a kind of full convolutional coding structure, it is on the basis of traditional coder-decoder framework, in coding module Jump is added between the respective layer (layer with an equal amount of characteristic pattern) of decoder module to link.
U-Net networks are divided into coding layer (totally eight layers), and decoding layer (totally eight layers) two parts often encode one layer, characteristic pattern (feature map) is long and width halves, and the feature number of plies increases half, often decodes one layer, characteristic pattern length and width double, the feature number of plies Increase doubles, i.e., is also concatenated by channel with corresponding coding layer, then carries out deconvolution processing.
Mirror image operation done to the surrounding of input picture, the quantitative design of convolutional layer is at 20,4 down-samplings, is adopted on 4 times Sample.Specifically, for n-layer network, the present invention adds between each i-th layer and the n-th-i layers and redirects connection, i-th layer and All channels in n-th-i layers are connected.
Based on the convolutional neural networks (CNN) for classification, design differentiates network model.Differentiate that the training order of network exists Before generating network, discrimination model is actually that can serve as its loss function to generate model, therefore arbiter is than generator Training ground for the convergence of generator more fully so that provide correct target.The network is designed to concatenate layer containing there are one, With four layers of convolutional layer.The CNN for reducing parameter designing, only divides the true and false property of each block in the blending image of generation Class, on the image convolution run the network, to it is all response do mean value, to provide the final output of D.
3. building the target detection model based on depth convolutional neural networks.Based on deep learning, the present invention passes through structure Neural network model with more hidden layers can be realized and learn more useful feature from large scale training data, to finally carry Rise the accuracy of classification or prediction.To realize the high-precision and high-adaptability of remote sensing target detection, herein, the present invention uses base The target detection network structure of plinth feature extraction network+Area generation network+sorter network is detected the structure of network model It builds.
Steps are as follows for the realization for the detection network algorithm based on depth network that the present invention designs：
(1) remote sensing images of the input after panchromatic sharpening；
(2) whole pictures are inputted into convolutional neural networks, carries out feature extraction；
(3) it is generated with RPN and suggests window, 300 suggestion windows are generated per pictures；
(4) suggestion window is mapped on last layer of convolution characteristic pattern of convolutional neural networks；
(5) each area-of-interest is made to generate fixed-size characteristic pattern by pond layer；
(6) it returns to return class probability and position using detection class probability and detection frame and carries out joint training.
Wherein, feature extraction network realizes network structure using residual error network structure (ResNet) by residual error network In-depth is obviously improved with classifying quality.Residual error network is reduced compared to traditional convolutional neural networks such as VGG complexities, is needed Parameter decline can accomplish deeper, the problem of being not in gradient disperse.Depth convolution residual error network be study be input to it is (defeated Go out-input) mapping, thus to obtain the prior information that is made of importation of output.
The residual error network for building one 18 layers and one 34 layers first, is inserted into shortcut on simple network, can be greatly Mitigate calculation amount.By introducing a shortcut connection between output input, rather than net is simply stacked in conventional method Network, to solve the problems, such as that gradient disappearance occurs in deep layer network.
Area generation network (Region Proposal Network, RPN), can be in the characteristic pattern that ResNet has been extracted On, currently relatively sparse all possible candidate frame is differentiated.Utilize the mapping mechanism of SPP-Net, Area generation net Network maps back artwork according to one-to-one point from convolutional layer, according to the different fixation initial gauges of design, to train network, root According to the accurate level of coverage with reference standard, its positive and negative label is given, enables whether its study the inside has target object.
In order to reduce the computation complexity of Area generation network, it is based on deep layer network, shared convolutional calculation knot may be implemented Fruit, fixed size variation, engineer's scale variation and sample mode, then obtains object candidate area, the i.e. candidate window of feature.It is first 9 kinds of candidate windows first are generated according to scale and length-width ratio, the window of fixed size is used on last layer of characteristic pattern of convolution Sliding, each window can export the feature of fixed size dimension, each window to 9 candidate targets return coordinate and Classification.
The object function of Area generation network be classification and return loss and.Classification uses cross entropy, returns using steady Fixed Smooth L1, formula are represented by：
Whole loss function is specially：
Loss function is divided into two parts, corresponds to error in classification whether two branches, i.e. target of Area generation network With the regression error of detection block, whereinUsing smooth L1 functions, study is more easily adjusted in the error than L2 form Rate.For writing for detection block, only consider to be determined as the candidate window for having target, and the coordinate marked is as the mesh write Mark.In addition, when calculating detection block error, the coordinate at four angles, t are not compared instead ofx, tY, tW, tH, as described below, specific four The calculation of a dimension：
tX=(x-xa)/wa,tY=(y-ya)/ha,
tW=log (w/wa),th=log (h/ha),
tX *=(x*-xa)/wa,tY *=(y*-ya)/ha,
tW *=log (w*/wa),th *=log (h*/ha),
In test, area-of-interest (ROI) pond layer obtains candidate ROI lists from Area generation network, passes through volume Lamination takes all features, carries out subsequent classification and recurrence.It is built by Area generation network and the public generation of detection network The convolutional layer for discussing window can be realized and generate candidate sharing between detection.
The training process of above-mentioned network uses four step coaching methods, the first step individually to train Area generation network, network parameter It is loaded by pre-training model；Second step, individually training detection network, the output candidate region of first step Area generation network is made To detect the input of network.Area generation network exports a candidate frame, intercepts original image by candidate frame, and will be after interception Image is operated by convolution several times-pondization, then exports two branches by ROI pondizations, is the detection classification of target classification respectively Probability (Softmax Loss) and detection frame return (Smooth L1 Loss).Third walks, and trains Area generation network again, The parameter of fixed network common portion at this time, update area generate the parameter of the exclusive part of network；Finally, according to Area generation The result of network finely tunes detection network structure, the parameter of fixed common portion again, and only update detects the exclusive part of network frame Parameter.
4. cascade network carries out end-to-end test.Panchromatic remote sensing images in input database test set, according to training Generation confrontation network panchromatic Edge contrast is carried out to it.It is then input in the detection network model built, it is detected As a result it is evaluated.For remote sensing images in database, in the test data set that invention content 1 is built, respectively according to fusion Front and back spectrum picture builds the data set based on PASCAL VOC contest formats, demonstrates the remote sensing figure after carrying out panchromatic sharpening As having better testing result；It is detected using traditional images treatment classification method and the depth detection network built herein Experiment, is detected the aircraft in figure, steamer, oil storage tank, bridge, five class the army and the people's target of harbour, is examined compared to traditional algorithm Effect is surveyed to be obviously improved.
Method of determination and evaluation is as follows：
It is ALL by all picture numbers of system testing, system identification goes out to have picture existing for five classes target to be detected Amount of images in set 1 is denoted as F, including identifying target without target originally and had target to know originally Incorrect picture number, is denoted as FP and FN, then F=FP+FN respectively；System identification is gone out into picture set 2 existing for no target In amount of images be denoted as T, including being identified correctly without target originally and had target without identifying mesh originally Target picture number is denoted as TP and TN, then T=TP+TN respectively.This system defines following finger according to actual identification needs Mark：

Claims (1)

1. a kind of Remote Sensing Target detection method based on deep learning, includes the following steps：
1) remote sensing images are utilized to build associated data set：After remote sensing images are classified and are marked, image data set and process The class label that markers work generates, and training set and test set are divided, it is used for subsequent network training and test；
2) it builds based on the panchromatic sharpening model for generating confrontation network：The generation network G of GAN be can learn from random noise to The image x in data set described in z and 1 is measured, to the mapping of the sample image y generated, i.e. G：{ x, z } → y generates model and takes The U-Net structures for redirecting connection are increased, coding layer and decoding layer two parts are divided into, often encode one layer, characteristic pattern length and width subtract Half, the feature number of plies increases half, often decodes one layer, the length and width of characteristic pattern double, and feature number of plies increase doubles and corresponding volume Code layer is concatenated by channel, then carries out deconvolution processing；Based on the convolutional neural networks CNN for classification, design differentiates net Network model, the network are designed to containing there are one concatenation layer and four layers of convolutional layers；
3) the target detection model based on depth convolutional neural networks is built：It is generated according to the candidate region of algorithm of target detection, Feature extraction, classification, four steps of position refine, within above-mentioned steps unification to a depth network frame, in GPU Concurrent operation, feature extraction is using residual error network ResNet as basic sorter network, wherein including several convolutional layers and linear list First ReLU, design section generate network structure and differentiate to all possible candidate frame on the characteristic pattern extracted, lead to Shared convolution is crossed, the marginal cost for calculating Suggestion box is reduced, model is carried out by backpropagation and stochastic gradient descent method End-to-end training；
4) end-to-end test is carried out to the model built：Based on the data set built in step 1, training objective detection model And test model.
CN201810235045.5A 2018-03-21 2018-03-21 Remote sensing image target detection method based on deep learning Active CN108564109B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810235045.5A CN108564109B (en) 2018-03-21 2018-03-21 Remote sensing image target detection method based on deep learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810235045.5A CN108564109B (en) 2018-03-21 2018-03-21 Remote sensing image target detection method based on deep learning

Publications (2)

Publication Number Publication Date
CN108564109A true CN108564109A (en) 2018-09-21
CN108564109B CN108564109B (en) 2021-08-10

Family

ID=63532037

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810235045.5A Active CN108564109B (en) 2018-03-21 2018-03-21 Remote sensing image target detection method based on deep learning

Country Status (1)

Country Link
CN (1) CN108564109B (en)

Cited By (61)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109344778A (en) * 2018-10-10 2019-02-15 成都信息工程大学 Based on the unmanned plane road extraction method for generating confrontation network
CN109377501A (en) * 2018-09-30 2019-02-22 上海鹰觉科技有限公司 Remote sensing images naval vessel dividing method and system based on transfer learning
CN109543740A (en) * 2018-11-14 2019-03-29 哈尔滨工程大学 A kind of object detection method based on generation confrontation network
CN109544563A (en) * 2018-11-12 2019-03-29 北京航空航天大学 A kind of passive millimeter wave image human body target dividing method towards violated object safety check
CN109727207A (en) * 2018-12-06 2019-05-07 华南理工大学 High spectrum image sharpening method based on Forecast of Spectra residual error convolutional neural networks
CN109740677A (en) * 2019-01-07 2019-05-10 湖北工业大学 It is a kind of to improve the semisupervised classification method for generating confrontation network based on principal component analysis
CN109859143A (en) * 2019-02-22 2019-06-07 中煤航测遥感集团有限公司 The panchromatic sharpening method of high spectrum image, device and electronic equipment
CN109859110A (en) * 2018-11-19 2019-06-07 华南理工大学 The panchromatic sharpening method of high spectrum image of control convolutional neural networks is tieed up based on spectrum
CN109858482A (en) * 2019-01-16 2019-06-07 创新奇智(重庆)科技有限公司 A kind of image key area detection method and its system, terminal device
CN109886114A (en) * 2019-01-18 2019-06-14 杭州电子科技大学 A kind of Ship Target Detection method based on cluster translation feature extraction strategy
CN109902602A (en) * 2019-02-16 2019-06-18 北京工业大学 A kind of airfield runway foreign materials recognition methods based on confrontation Neural Network Data enhancing
CN109919123A (en) * 2019-03-19 2019-06-21 自然资源部第一海洋研究所 Offshore spilled oil detection method based on Analysis On Multi-scale Features depth convolutional neural networks
CN109949299A (en) * 2019-03-25 2019-06-28 东南大学 A kind of cardiologic medical image automatic segmentation method
CN109948469A (en) * 2019-03-01 2019-06-28 吉林大学 The automatic detection recognition method of crusing robot instrument based on deep learning
CN110110576A (en) * 2019-01-03 2019-08-09 北京航空航天大学 A kind of traffic scene thermal infrared semanteme generation method based on twin semantic network
CN110175548A (en) * 2019-05-20 2019-08-27 中国科学院光电技术研究所 Remote sensing images building extracting method based on attention mechanism and channel information
CN110197147A (en) * 2019-05-23 2019-09-03 星际空间（天津）科技发展有限公司 Building Cass collection method, apparatus, storage medium and the equipment of remote sensing image
CN110222622A (en) * 2019-05-31 2019-09-10 甘肃省祁连山水源涵养林研究院 A kind of ambient soil detection method and device
CN110427793A (en) * 2019-08-01 2019-11-08 厦门商集网络科技有限责任公司 A kind of code detection method and its system based on deep learning
CN110472699A (en) * 2019-08-24 2019-11-19 福州大学 A kind of harmful biological motion blurred picture detection method of field of electric force institute based on GAN
CN110473144A (en) * 2019-08-07 2019-11-19 南京信息工程大学 A kind of image super-resolution rebuilding method based on laplacian pyramid network
CN110502654A (en) * 2019-08-26 2019-11-26 长光卫星技术有限公司 A kind of object library generation system suitable for multi-source heterogeneous remotely-sensed data
CN110751699A (en) * 2019-10-15 2020-02-04 西安电子科技大学 Color reconstruction method of optical remote sensing image based on convolutional neural network
CN110763698A (en) * 2019-10-12 2020-02-07 仲恺农业工程学院 Hyperspectral citrus leaf disease identification method based on characteristic wavelength
CN110782398A (en) * 2018-12-13 2020-02-11 北京嘀嘀无限科技发展有限公司 Image processing method, generation type countermeasure network, electronic equipment and storage medium
CN110929618A (en) * 2019-11-15 2020-03-27 国网江西省电力有限公司电力科学研究院 Potential safety hazard detection and evaluation method for power distribution network crossing type building
CN111026899A (en) * 2019-12-11 2020-04-17 兰州理工大学 Product generation method based on deep learning
CN111027509A (en) * 2019-12-23 2020-04-17 武汉大学 Hyperspectral image target detection method based on double-current convolution neural network
CN111104840A (en) * 2018-10-26 2020-05-05 斯特拉德视觉公司 Learning and testing method and device based on regression loss
CN111127416A (en) * 2019-12-19 2020-05-08 武汉珈鹰智能科技有限公司 Computer vision-based automatic detection method for surface defects of concrete structure
CN111161250A (en) * 2019-12-31 2020-05-15 北京云智宇航科技有限公司 Multi-scale remote sensing image dense house detection method and device
CN111210483A (en) * 2019-12-23 2020-05-29 中国人民解放军空军研究院战场环境研究所 Simulated satellite cloud picture generation method based on generation of countermeasure network and numerical mode product
CN111259710A (en) * 2018-12-03 2020-06-09 初速度（苏州）科技有限公司 Parking space structure detection model training method adopting parking space frame lines and end points
CN111274865A (en) * 2019-12-14 2020-06-12 深圳先进技术研究院 Remote sensing image cloud detection method and device based on full convolution neural network
CN111368843A (en) * 2020-03-06 2020-07-03 电子科技大学 Method for extracting lake on ice based on semantic segmentation
CN111428678A (en) * 2020-04-02 2020-07-17 山东卓元数据技术有限公司 Method for generating confrontation network remote sensing image sample expansion under space constraint condition for ground object change detection
CN111488782A (en) * 2019-01-25 2020-08-04 斯特拉德视觉公司 Method and device for detecting road barrier by using Laplacian pyramid network
CN111488893A (en) * 2019-01-25 2020-08-04 银河水滴科技（北京）有限公司 Image classification method and device
US10748033B2 (en) 2018-12-11 2020-08-18 Industrial Technology Research Institute Object detection method using CNN model and object detection apparatus using the same
CN111553212A (en) * 2020-04-16 2020-08-18 中国科学院深圳先进技术研究院 Remote sensing image target detection method based on smooth frame regression function
WO2020164270A1 (en) * 2019-02-15 2020-08-20 平安科技（深圳）有限公司 Deep-learning-based pedestrian detection method, system and apparatus, and storage medium
CN111612066A (en) * 2020-05-21 2020-09-01 成都理工大学 Remote sensing image classification method based on depth fusion convolutional neural network
CN111611968A (en) * 2020-05-29 2020-09-01 中国科学院西北生态环境资源研究院 Processing method of remote sensing image and remote sensing image processing model
CN111696066A (en) * 2020-06-13 2020-09-22 中北大学 Multi-band image synchronous fusion and enhancement method based on improved WGAN-GP
CN111798662A (en) * 2020-07-31 2020-10-20 公安部交通管理科学研究所 Urban traffic accident early warning method based on space-time gridding data
CN111796310A (en) * 2020-07-02 2020-10-20 武汉北斗星度科技有限公司 High-precision positioning method, device and system based on Beidou GNSS
CN111832404A (en) * 2020-06-04 2020-10-27 中国科学院空天信息创新研究院 Small sample remote sensing ground feature classification method and system based on feature generation network
CN112100908A (en) * 2020-08-31 2020-12-18 西安工程大学 Garment design method for generating confrontation network based on multi-condition deep convolution
CN112184554A (en) * 2020-10-13 2021-01-05 重庆邮电大学 Remote sensing image fusion method based on residual mixed expansion convolution
CN112529114A (en) * 2021-01-13 2021-03-19 北京云真信科技有限公司 Target information identification method based on GAN, electronic device and medium
CN112926383A (en) * 2021-01-08 2021-06-08 浙江大学 Automatic target identification system based on underwater laser image
CN113012069A (en) * 2021-03-17 2021-06-22 中国科学院西安光学精密机械研究所 Optical remote sensing image quality improving method combining deep learning under wavelet transform domain
TWI740565B (en) * 2020-07-03 2021-09-21 財團法人國家實驗研究院國家高速網路與計算中心 Method for improving remote sensing image quality, computer program product and system thereof
WO2021184891A1 (en) * 2020-03-20 2021-09-23 中国科学院深圳先进技术研究院 Remotely-sensed image-based terrain classification method, and system
CN114241274A (en) * 2021-11-30 2022-03-25 电子科技大学 Small target detection method based on super-resolution multi-scale feature fusion
CN114241339A (en) * 2022-02-28 2022-03-25 山东力聚机器人科技股份有限公司 Remote sensing image recognition model, method and system, server and medium
CN114973021A (en) * 2022-06-15 2022-08-30 北京鹏鹄物宇科技发展有限公司 Satellite image data processing system and method based on deep learning
CN115512098A (en) * 2022-09-26 2022-12-23 重庆大学 Electronic bridge inspection system and inspection method
CN115689947A (en) * 2022-12-30 2023-02-03 杭州魔点科技有限公司 Image sharpening method, system, electronic device and storage medium
CN117237334A (en) * 2023-11-09 2023-12-15 江西联益光学有限公司 Deep learning-based method for detecting stray light of mobile phone lens
CN118262245A (en) * 2024-05-28 2024-06-28 山东锋士信息技术有限公司 River and lake management violation problem remote sensing monitoring method based on Laplace and similarity

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120269430A1 (en) * 2011-04-22 2012-10-25 Michael Paul Deskevich System and method for combining color information with spatial information in multispectral images
US20130129201A1 (en) * 2011-11-18 2013-05-23 Dehong Liu Method for Pan-Sharpening Panchromatic and Multispectral Images Using Wavelet Dictionaries
CN104268847A (en) * 2014-09-23 2015-01-07 西安电子科技大学 Infrared light image and visible light image fusion method based on interactive non-local average filtering

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120269430A1 (en) * 2011-04-22 2012-10-25 Michael Paul Deskevich System and method for combining color information with spatial information in multispectral images
US20130129201A1 (en) * 2011-11-18 2013-05-23 Dehong Liu Method for Pan-Sharpening Panchromatic and Multispectral Images Using Wavelet Dictionaries
CN104268847A (en) * 2014-09-23 2015-01-07 西安电子科技大学 Infrared light image and visible light image fusion method based on interactive non-local average filtering

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
LIUYUNHONG，ET AL: "《Remote sensing image fusion based on two-stream fusion network》", 《INTERNATIONAL CONFERENCE ON MULTIMEDIA MODELING》 *

Cited By (95)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109377501A (en) * 2018-09-30 2019-02-22 上海鹰觉科技有限公司 Remote sensing images naval vessel dividing method and system based on transfer learning
CN109344778A (en) * 2018-10-10 2019-02-15 成都信息工程大学 Based on the unmanned plane road extraction method for generating confrontation network
CN111104840B (en) * 2018-10-26 2024-01-09 斯特拉德视觉公司 Regression loss-based learning and testing method and device
CN111104840A (en) * 2018-10-26 2020-05-05 斯特拉德视觉公司 Learning and testing method and device based on regression loss
CN109544563A (en) * 2018-11-12 2019-03-29 北京航空航天大学 A kind of passive millimeter wave image human body target dividing method towards violated object safety check
CN109543740B (en) * 2018-11-14 2022-07-15 哈尔滨工程大学 Target detection method based on generation countermeasure network
CN109543740A (en) * 2018-11-14 2019-03-29 哈尔滨工程大学 A kind of object detection method based on generation confrontation network
CN109859110B (en) * 2018-11-19 2023-01-06 华南理工大学 Hyperspectral image panchromatic sharpening method based on spectrum dimension control convolutional neural network
CN109859110A (en) * 2018-11-19 2019-06-07 华南理工大学 The panchromatic sharpening method of high spectrum image of control convolutional neural networks is tieed up based on spectrum
CN111259710B (en) * 2018-12-03 2022-06-10 魔门塔(苏州)科技有限公司 Parking space structure detection model training method adopting parking space frame lines and end points
CN111259710A (en) * 2018-12-03 2020-06-09 初速度（苏州）科技有限公司 Parking space structure detection model training method adopting parking space frame lines and end points
CN109727207B (en) * 2018-12-06 2022-12-16 华南理工大学 Hyperspectral image sharpening method based on spectrum prediction residual convolution neural network
CN109727207A (en) * 2018-12-06 2019-05-07 华南理工大学 High spectrum image sharpening method based on Forecast of Spectra residual error convolutional neural networks
US10748033B2 (en) 2018-12-11 2020-08-18 Industrial Technology Research Institute Object detection method using CNN model and object detection apparatus using the same
TWI708209B (en) * 2018-12-11 2020-10-21 財團法人工業技術研究院 Object detection method using cnn model and object detection apparatus using the same
CN110782398B (en) * 2018-12-13 2020-12-18 北京嘀嘀无限科技发展有限公司 Image processing method, generative countermeasure network system and electronic device
CN110782398A (en) * 2018-12-13 2020-02-11 北京嘀嘀无限科技发展有限公司 Image processing method, generation type countermeasure network, electronic equipment and storage medium
CN110110576A (en) * 2019-01-03 2019-08-09 北京航空航天大学 A kind of traffic scene thermal infrared semanteme generation method based on twin semantic network
CN109740677A (en) * 2019-01-07 2019-05-10 湖北工业大学 It is a kind of to improve the semisupervised classification method for generating confrontation network based on principal component analysis
CN109858482A (en) * 2019-01-16 2019-06-07 创新奇智(重庆)科技有限公司 A kind of image key area detection method and its system, terminal device
CN109858482B (en) * 2019-01-16 2020-04-14 创新奇智(重庆)科技有限公司 Image key area detection method and system and terminal equipment
CN109886114A (en) * 2019-01-18 2019-06-14 杭州电子科技大学 A kind of Ship Target Detection method based on cluster translation feature extraction strategy
CN111488782A (en) * 2019-01-25 2020-08-04 斯特拉德视觉公司 Method and device for detecting road barrier by using Laplacian pyramid network
CN111488782B (en) * 2019-01-25 2023-04-07 斯特拉德视觉公司 Method and device for detecting road barrier by using Laplacian pyramid network
CN111488893A (en) * 2019-01-25 2020-08-04 银河水滴科技（北京）有限公司 Image classification method and device
CN111488893B (en) * 2019-01-25 2023-05-30 银河水滴科技（北京）有限公司 Image classification method and device
WO2020164270A1 (en) * 2019-02-15 2020-08-20 平安科技（深圳）有限公司 Deep-learning-based pedestrian detection method, system and apparatus, and storage medium
CN109902602A (en) * 2019-02-16 2019-06-18 北京工业大学 A kind of airfield runway foreign materials recognition methods based on confrontation Neural Network Data enhancing
CN109902602B (en) * 2019-02-16 2021-04-30 北京工业大学 Method for identifying foreign matter material of airport runway based on antagonistic neural network data enhancement
CN109859143B (en) * 2019-02-22 2020-09-29 中煤航测遥感集团有限公司 Hyperspectral image panchromatic sharpening method and device and electronic equipment
CN109859143A (en) * 2019-02-22 2019-06-07 中煤航测遥感集团有限公司 The panchromatic sharpening method of high spectrum image, device and electronic equipment
CN109948469B (en) * 2019-03-01 2022-11-29 吉林大学 Automatic inspection robot instrument detection and identification method based on deep learning
CN109948469A (en) * 2019-03-01 2019-06-28 吉林大学 The automatic detection recognition method of crusing robot instrument based on deep learning
CN109919123A (en) * 2019-03-19 2019-06-21 自然资源部第一海洋研究所 Offshore spilled oil detection method based on Analysis On Multi-scale Features depth convolutional neural networks
CN109949299A (en) * 2019-03-25 2019-06-28 东南大学 A kind of cardiologic medical image automatic segmentation method
CN110175548B (en) * 2019-05-20 2022-08-23 中国科学院光电技术研究所 Remote sensing image building extraction method based on attention mechanism and channel information
CN110175548A (en) * 2019-05-20 2019-08-27 中国科学院光电技术研究所 Remote sensing images building extracting method based on attention mechanism and channel information
CN110197147B (en) * 2019-05-23 2022-12-02 星际空间（天津）科技发展有限公司 Building example extraction method, device, storage medium and equipment of remote sensing image
CN110197147A (en) * 2019-05-23 2019-09-03 星际空间（天津）科技发展有限公司 Building Cass collection method, apparatus, storage medium and the equipment of remote sensing image
CN110222622A (en) * 2019-05-31 2019-09-10 甘肃省祁连山水源涵养林研究院 A kind of ambient soil detection method and device
CN110427793A (en) * 2019-08-01 2019-11-08 厦门商集网络科技有限责任公司 A kind of code detection method and its system based on deep learning
CN110427793B (en) * 2019-08-01 2022-04-26 厦门商集网络科技有限责任公司 Bar code detection method and system based on deep learning
CN110473144A (en) * 2019-08-07 2019-11-19 南京信息工程大学 A kind of image super-resolution rebuilding method based on laplacian pyramid network
CN110473144B (en) * 2019-08-07 2023-04-25 南京信息工程大学 Image super-resolution reconstruction method based on Laplacian pyramid network
CN110472699A (en) * 2019-08-24 2019-11-19 福州大学 A kind of harmful biological motion blurred picture detection method of field of electric force institute based on GAN
CN110502654A (en) * 2019-08-26 2019-11-26 长光卫星技术有限公司 A kind of object library generation system suitable for multi-source heterogeneous remotely-sensed data
CN110763698A (en) * 2019-10-12 2020-02-07 仲恺农业工程学院 Hyperspectral citrus leaf disease identification method based on characteristic wavelength
CN110751699A (en) * 2019-10-15 2020-02-04 西安电子科技大学 Color reconstruction method of optical remote sensing image based on convolutional neural network
CN110751699B (en) * 2019-10-15 2023-03-10 西安电子科技大学 Color reconstruction method of optical remote sensing image based on convolutional neural network
CN110929618A (en) * 2019-11-15 2020-03-27 国网江西省电力有限公司电力科学研究院 Potential safety hazard detection and evaluation method for power distribution network crossing type building
CN110929618B (en) * 2019-11-15 2023-06-20 国网江西省电力有限公司电力科学研究院 Potential safety hazard detection and assessment method for power distribution network crossing type building
CN111026899A (en) * 2019-12-11 2020-04-17 兰州理工大学 Product generation method based on deep learning
CN111274865A (en) * 2019-12-14 2020-06-12 深圳先进技术研究院 Remote sensing image cloud detection method and device based on full convolution neural network
CN111274865B (en) * 2019-12-14 2023-09-19 深圳先进技术研究院 Remote sensing image cloud detection method and device based on full convolution neural network
CN111127416A (en) * 2019-12-19 2020-05-08 武汉珈鹰智能科技有限公司 Computer vision-based automatic detection method for surface defects of concrete structure
CN111210483A (en) * 2019-12-23 2020-05-29 中国人民解放军空军研究院战场环境研究所 Simulated satellite cloud picture generation method based on generation of countermeasure network and numerical mode product
CN111027509A (en) * 2019-12-23 2020-04-17 武汉大学 Hyperspectral image target detection method based on double-current convolution neural network
CN111161250B (en) * 2019-12-31 2023-05-26 南遥科技(广东)有限公司 Method and device for detecting dense houses by using multi-scale remote sensing images
CN111161250A (en) * 2019-12-31 2020-05-15 北京云智宇航科技有限公司 Multi-scale remote sensing image dense house detection method and device
CN111368843B (en) * 2020-03-06 2022-06-10 电子科技大学 Method for extracting lake on ice based on semantic segmentation
CN111368843A (en) * 2020-03-06 2020-07-03 电子科技大学 Method for extracting lake on ice based on semantic segmentation
WO2021184891A1 (en) * 2020-03-20 2021-09-23 中国科学院深圳先进技术研究院 Remotely-sensed image-based terrain classification method, and system
CN111428678A (en) * 2020-04-02 2020-07-17 山东卓元数据技术有限公司 Method for generating confrontation network remote sensing image sample expansion under space constraint condition for ground object change detection
CN111428678B (en) * 2020-04-02 2023-06-23 山东卓智软件股份有限公司 Method for generating remote sensing image sample expansion of countermeasure network under space constraint condition
WO2021208502A1 (en) * 2020-04-16 2021-10-21 中国科学院深圳先进技术研究院 Remote-sensing image target detection method based on smooth bounding box regression function
CN111553212A (en) * 2020-04-16 2020-08-18 中国科学院深圳先进技术研究院 Remote sensing image target detection method based on smooth frame regression function
CN111612066B (en) * 2020-05-21 2022-03-08 成都理工大学 Remote sensing image classification method based on depth fusion convolutional neural network
CN111612066A (en) * 2020-05-21 2020-09-01 成都理工大学 Remote sensing image classification method based on depth fusion convolutional neural network
CN111611968B (en) * 2020-05-29 2022-02-01 中国科学院西北生态环境资源研究院 Processing method of remote sensing image and remote sensing image processing model
CN111611968A (en) * 2020-05-29 2020-09-01 中国科学院西北生态环境资源研究院 Processing method of remote sensing image and remote sensing image processing model
CN111832404A (en) * 2020-06-04 2020-10-27 中国科学院空天信息创新研究院 Small sample remote sensing ground feature classification method and system based on feature generation network
CN111696066A (en) * 2020-06-13 2020-09-22 中北大学 Multi-band image synchronous fusion and enhancement method based on improved WGAN-GP
CN111796310A (en) * 2020-07-02 2020-10-20 武汉北斗星度科技有限公司 High-precision positioning method, device and system based on Beidou GNSS
CN111796310B (en) * 2020-07-02 2024-02-02 武汉北斗星度科技有限公司 Beidou GNSS-based high-precision positioning method, device and system
TWI740565B (en) * 2020-07-03 2021-09-21 財團法人國家實驗研究院國家高速網路與計算中心 Method for improving remote sensing image quality, computer program product and system thereof
CN111798662A (en) * 2020-07-31 2020-10-20 公安部交通管理科学研究所 Urban traffic accident early warning method based on space-time gridding data
CN112100908B (en) * 2020-08-31 2024-03-22 西安工程大学 Clothing design method for generating countermeasure network based on multi-condition deep convolution
CN112100908A (en) * 2020-08-31 2020-12-18 西安工程大学 Garment design method for generating confrontation network based on multi-condition deep convolution
CN112184554B (en) * 2020-10-13 2022-08-23 重庆邮电大学 Remote sensing image fusion method based on residual mixed expansion convolution
CN112184554A (en) * 2020-10-13 2021-01-05 重庆邮电大学 Remote sensing image fusion method based on residual mixed expansion convolution
CN112926383A (en) * 2021-01-08 2021-06-08 浙江大学 Automatic target identification system based on underwater laser image
CN112926383B (en) * 2021-01-08 2023-03-03 浙江大学 Automatic target identification system based on underwater laser image
CN112529114A (en) * 2021-01-13 2021-03-19 北京云真信科技有限公司 Target information identification method based on GAN, electronic device and medium
CN113012069B (en) * 2021-03-17 2023-09-05 中国科学院西安光学精密机械研究所 Optical remote sensing image quality improving method combining deep learning under wavelet transform domain
CN113012069A (en) * 2021-03-17 2021-06-22 中国科学院西安光学精密机械研究所 Optical remote sensing image quality improving method combining deep learning under wavelet transform domain
CN114241274B (en) * 2021-11-30 2023-04-07 电子科技大学 Small target detection method based on super-resolution multi-scale feature fusion
CN114241274A (en) * 2021-11-30 2022-03-25 电子科技大学 Small target detection method based on super-resolution multi-scale feature fusion
CN114241339A (en) * 2022-02-28 2022-03-25 山东力聚机器人科技股份有限公司 Remote sensing image recognition model, method and system, server and medium
CN114973021A (en) * 2022-06-15 2022-08-30 北京鹏鹄物宇科技发展有限公司 Satellite image data processing system and method based on deep learning
CN115512098B (en) * 2022-09-26 2023-09-01 重庆大学 Bridge electronic inspection system and inspection method
CN115512098A (en) * 2022-09-26 2022-12-23 重庆大学 Electronic bridge inspection system and inspection method
CN115689947A (en) * 2022-12-30 2023-02-03 杭州魔点科技有限公司 Image sharpening method, system, electronic device and storage medium
CN117237334A (en) * 2023-11-09 2023-12-15 江西联益光学有限公司 Deep learning-based method for detecting stray light of mobile phone lens
CN117237334B (en) * 2023-11-09 2024-03-26 江西联益光学有限公司 Deep learning-based method for detecting stray light of mobile phone lens
CN118262245A (en) * 2024-05-28 2024-06-28 山东锋士信息技术有限公司 River and lake management violation problem remote sensing monitoring method based on Laplace and similarity

Also Published As

Publication number Publication date
CN108564109B (en) 2021-08-10

Similar Documents

Publication Publication Date Title
CN108564109A (en) A kind of Remote Sensing Target detection method based on deep learning
CN110363215B (en) Method for converting SAR image into optical image based on generating type countermeasure network
CN106815601B (en) Hyperspectral image classification method based on recurrent neural network
CN108537742A (en) A kind of panchromatic sharpening method of remote sensing images based on generation confrontation network
CN105787439B (en) A kind of depth image human synovial localization method based on convolutional neural networks
CN110135267A (en) A kind of subtle object detection method of large scene SAR image
CN111368896A (en) Hyperspectral remote sensing image classification method based on dense residual three-dimensional convolutional neural network
CN112070078B (en) Deep learning-based land utilization classification method and system
CN105809194B (en) A kind of method that SAR image is translated as optical image
CN109145992A (en) Cooperation generates confrontation network and sky composes united hyperspectral image classification method
CN109993072A (en) The low resolution pedestrian weight identifying system and method generated based on super resolution image
CN111625608A (en) Method and system for generating electronic map according to remote sensing image based on GAN model
CN103268607B (en) A kind of common object detection method under weak supervision condition
CN108229551B (en) Hyperspectral remote sensing image classification method based on compact dictionary sparse representation
CN113936214B (en) Karst wetland vegetation community classification method based on fusion of aerospace remote sensing images
CN115131313A (en) Hyperspectral image change detection method and device based on Transformer
CN112949380B (en) Intelligent underwater target identification system based on laser radar point cloud data
CN114220007A (en) Hyperspectral image band selection method based on overcomplete depth low-rank subspace clustering
CN113724308B (en) Cross-waveband stereo matching algorithm based on mutual attention of luminosity and contrast
CN112529828A (en) Reference data non-sensitive remote sensing image space-time fusion model construction method
CN117541665A (en) Infrared image generation method based on improved generation countermeasure network
CN117392533A (en) Construction method of high-spectrum remote sensing image lithology sample set
Song et al. Hyperspectral image classification using hierarchical spatial-spectral transformer
CN116452965A (en) Underwater target detection and recognition method based on acousto-optic fusion
CN115719454A (en) Remote sensing image scene matching method combined with view angle conversion

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant