CN117853932B - Sea surface target detection method, detection platform and system based on photoelectric pod - Google Patents
Sea surface target detection method, detection platform and system based on photoelectric pod Download PDFInfo
- Publication number
- CN117853932B CN117853932B CN202410248872.3A CN202410248872A CN117853932B CN 117853932 B CN117853932 B CN 117853932B CN 202410248872 A CN202410248872 A CN 202410248872A CN 117853932 B CN117853932 B CN 117853932B
- Authority
- CN
- China
- Prior art keywords
- sea surface
- image
- image data
- target detection
- surface target
- 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.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 126
- 230000005693 optoelectronics Effects 0.000 claims abstract description 22
- 230000001629 suppression Effects 0.000 claims abstract description 12
- 230000009466 transformation Effects 0.000 claims abstract description 7
- 238000000034 method Methods 0.000 claims description 10
- 238000004422 calculation algorithm Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 6
- 230000004927 fusion Effects 0.000 claims description 5
- 238000005259 measurement Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 description 7
- 230000002776 aggregation Effects 0.000 description 5
- 238000004220 aggregation Methods 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000003044 adaptive effect Effects 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 241000251468 Actinopterygii Species 0.000 description 1
- 206010039203 Road traffic accident Diseases 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000009429 distress Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000035515 penetration Effects 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- 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
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a sea surface target detection method, a detection platform and a system based on a photoelectric pod, which belong to the field of sea surface target detection and comprise the following steps: for the aggregated multiband sea surface image data transmitted by the optoelectronic pod, performing: depolymerizing it into a plurality of single-pass image data; respectively executing sea clutter self-adaptive suppression steps on each single-path image data to obtain a denoised image; respectively carrying out target detection on the denoised images under the same focal length, and fusing detection results; the sea clutter self-adaptive suppression step comprises the following steps: performing top hat transformation on the image data to be processed, calculating gradients of the image data in different directions, and taking the minimum pixel value of a plurality of gradient images at the same position as the pixel value of the denoised image at the position; and adjusting the pose of the tele lens by using the detection result under the wide-angle lens. The invention can fully utilize the multichannel sea surface image data acquired by the photoelectric pod, inhibit sea surface clutter therein and improve the detection precision of sea surface targets.
Description
Technical Field
The invention belongs to the field of sea surface target detection, and particularly relates to a sea surface target detection method, a detection platform and a system based on a photoelectric pod.
Background
On one hand, with the continuous expansion and penetration of weather factors such as typhoons and activities such as ocean oil gas development and the like, the number of water traffic accidents is in an ascending trend, and serious economic loss and casualties of personnel are extremely easy to be caused in a severe offshore environment, so that the rapid identification and tracking of offshore distress targets are important to ensure a safe navigation environment. On the other hand, the marine protection significance is great, the marine initiative can be better mastered only by rapidly positioning and monitoring and hitting an invading ship, so that the complex sea surface environment is required to be monitored in a high-performance and all-round manner, and the intelligent detection of the marine target is a key for protecting the marine protection safety.
The problem of complex background usually exists in offshore target detection, and complex clutter in the sea background not only has high-brightness sea bright stripes and dense sea fish scale bright spots, but also has background clutter noise such as sea heterogeneous complex background and island clutter, so that targets are confused with complex sea clutter background noise, and the detection difficulty of sea targets is increased.
Along with miniaturization and diversification of short wave infrared, visible light, medium wave infrared, long wave infrared and other photoelectric detectors, image data of different wave bands can be acquired more conveniently. The information in the images of different wave bands also has some differences, such as the penetrability of infrared light, so that the infrared image can distinguish the target from the background according to the radiation difference, therefore, the infrared image can acquire more effective information at night, and the visible light can acquire texture and color information which are more in line with the human visual system. The selection of the lens is also richer and diversified: the wide-angle lens has the characteristics of short focal length and large visual angle. More scenes can be captured, and the long-focus lens has long focal length and small visual angle, so that distant targets can be clearly imaged. For the combination of the detectors with different wavebands and different lenses, how to process the image data of the detector is important to more efficiently utilize the characteristics of the multichannel image so as to obtain more accurate detection and identification results.
The photoelectric pod can be flexibly configured according to task requirements, and can be configured by different combinations of short-wave infrared, visible light and long-wave infrared detectors and long-focus and wide-angle lenses respectively to acquire sea surface image data with different wave bands and different focal lengths. In the patent document with the application publication number of CN116248705A, a micro photoelectric pod multichannel image transmission and processing system is disclosed, as shown in FIG. 1, the system comprises a plurality of photoelectric detectors, an FPGA unit A and an image aggregation unit, wherein the photoelectric detectors are used for collecting photon signals in different wave bands, generating image electric signals through a photoelectric conversion circuit, and outputting the image signals in the plurality of paths to the FPGA unit A; the FPGA unit A is used for realizing the acquisition of multiple paths of image signals, generating multiple paths of image data and outputting the multiple paths of image data to the image aggregation unit; the image aggregation unit aggregates the multipath image data into aggregated image data and outputs the aggregated image data to a single-channel high-speed link which is connected with the miniature photoelectric pod and a remote data processing platform (such as an airplane platform), so that synchronous real-time transmission, processing, forwarding or storage of the multipath detector image data is realized, and the method has the characteristics of good instantaneity, high transmission rate, low link cost, good flexibility and the like. The micro photoelectric pod is used for collecting sea surface image data, which is beneficial to improving the sea surface target detection effect, but how to effectively process the multipath image data to accurately finish sea surface target detection is still lacking in an effective method.
Disclosure of Invention
Aiming at the defects and improvement demands of the prior art, the invention provides a sea surface target detection method, a detection platform and a system based on an optoelectronic pod, which aim to fully utilize multichannel sea surface image data acquired by the optoelectronic pod and inhibit sea surface clutter therein so as to improve the detection precision of a sea surface target.
To achieve the above object, according to one aspect of the present invention, there is provided a sea surface target detection method based on a photoelectric pod, comprising:
Continuously receiving sea surface image data transmitted by the photoelectric pod; for each received frame of aggregated multi-band sea surface image data, performing the steps of:
(S1) depolymerizing it into a plurality of single-pass image data; each single-path image data corresponds to image data of a wave band under one focal length collected by the optoelectronic pod;
(S2) respectively executing sea clutter self-adaptive suppression steps on each single-path image data to obtain a corresponding denoised image; the sea clutter self-adaptive suppression step comprises the following steps:
(S21) performing top hat transformation on the image data to be processed to obtain a preliminary denoising image;
(S22) calculating gradients of the preliminary denoising image in different directions to obtain a plurality of gradient images;
(S23) taking the minimum pixel value of the plurality of gradient images at the same position as the pixel value of the denoised image at the position to obtain a denoised image corresponding to the image to be processed;
and (S3) respectively carrying out target detection on the denoised images under the same focal length, fusing detection results, and taking the fusion result as a sea surface target detection result under the corresponding focal length.
Further, the step (S22) includes:
carrying out connected domain marking on the preliminary denoising image, removing connected domains smaller than a preset threshold value, and calculating the average size C of the remaining connected domains;
Calculating a gradient convolution kernel size ks according to ks=c// M, and determining gradient convolution kernels in all directions according to the calculated gradient convolution kernel size;
Performing convolution operation on the preliminary denoising image by utilizing gradient convolution check under each direction to calculate gradients of the preliminary denoising image in different directions, so as to obtain a plurality of gradient images;
wherein "//" denotes integer division operations; m is a preset positive integer.
Further, m=5.
Further, the sea surface target detection method based on the photoelectric pod provided by the invention further comprises the following steps after the step (S3):
(S4) calculating the central position of the sea surface target according to the sea surface target detection result under the shorter focal length, and calculating the offset of the central position of the sea surface target relative to the center of the image;
And (S5) sending a command to the optoelectronic pod so that the optoelectronic pod adjusts the pose of the lens with a longer focal length according to the offset, and the central position of the sea surface target under the focal length is overlapped with the central position of the image.
Further, in step (S3), fusing the target detection results under the same focal length, including:
Calculating the overlapping degree between detection frames obtained by target detection of each denoised image under the focal length, outputting the detection frames with the overlapping degree larger than a preset high threshold as a trusted target, and rejecting the detection frames with the overlapping degree smaller than a preset low threshold as false alarms;
Determining the track of the sea surface target according to the sea surface target detection result sequence of the sea surface image data before the current frame, and outputting the detection frame with the overlapping degree between a low threshold value and a high threshold value as a trusted target if the detection frame is positioned on the track of the sea surface target, otherwise, rejecting the detection frame as a false alarm.
Further, in the step (S3), if the denoised image is a visible light band image, performing object detection by using YOLOv; and if the denoised image is an infrared band image, detecting by adopting a multi-scale block contrast measurement algorithm.
According to yet another aspect of the present invention, there is provided a photoelectric pod-based sea surface target detection platform comprising:
a computer readable storage medium storing a computer program;
and a processor for reading a computer program stored in a computer readable storage medium and executing the sea surface target detection method based on the photoelectric pod.
According to yet another aspect of the present invention there is provided a sea surface target detection system comprising:
The sea surface target detection platform based on the photoelectric pod provided by the invention;
an optoelectronic pod;
The image data stream is transmitted between the photoelectric pod and the sea surface target detection platform through a single channel link.
In general, by the above technical solutions conceived by the present invention, the following advantageous effects can be obtained.
(1) According to the sea surface image data collected by the photoelectric pod, top hat transformation is firstly carried out on the sea surface image data, sea surface scale waves, bright strips and other sea clutter noise with uniform brightness and simple texture on the sea surface can be effectively filtered, then the specific value of each pixel in the image is determined based on directional gradient calculation, so that noise which is close to the target size but inconsistent in shape in the image can be effectively restrained, sea clutter noise in the sea surface image data can be effectively restrained by combining the sea surface scale waves and the bright strips, on the basis, target detection is carried out on denoised multipath image data respectively, and the target detection results of images with different wave bands under the same focal length are used as sea surface target detection results under the corresponding focal length, so that the information of the images with different wave bands and different focal lengths can be fully utilized, and the sea surface target detection accuracy is effectively improved.
(2) In the preferable scheme of the invention, when noise close to the target size but inconsistent in shape in the image is restrained based on directional gradient calculation, the image is firstly marked with the connected domain, the connected domain with undersize is filtered, and then the gradient convolution kernel size for calculating the gradient is determined based on the average size of the rest connected domains, so that the size of the sea surface target to be detected can be self-adapted under different scenes, and the precision and the robustness of sea surface target detection are effectively improved. In a further preferred scheme of the invention, the gradient convolution kernel is further set as a result of dividing the average size of the connected domain by 5, and experimental data show that the gradient is calculated based on the size of the gradient convolution kernel, so that sea surface background clutter can be inhibited to the greatest extent.
(3) Under the shorter focal length, the view field is larger, noise and interference objects are more, the target size is small, the detection difficulty is high, under the longer focal length, the view field is smaller, the noise and interference objects are smaller, the target is clear, and the detection difficulty is low, but the sea surface target is possibly located outside the view field.
Drawings
FIG. 1 is a schematic diagram of a data connection between a conventional optoelectronic pod and an aircraft platform.
Fig. 2 is a schematic diagram of a sea surface target detection method based on a photoelectric pod according to an embodiment of the present invention.
Fig. 3 is an original image before top hat transformation according to an embodiment of the present invention.
Fig. 4 is a top hat transformed image of the original image of fig. 3.
Fig. 5 is a schematic diagram of gradient convolution kernels in 8 directions according to an embodiment of the present disclosure.
Fig. 6 is a gradient plot of the image of fig. 4 in the 0 deg. direction.
Fig. 7 is a gradient plot of the image of fig. 4 in the 90 direction.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention. In addition, the technical features of the embodiments of the present invention described below may be combined with each other as long as they do not collide with each other.
In the present invention, the terms "first," "second," and the like in the description and in the drawings, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order.
In order to effectively improve the detection precision of sea surface targets, the invention provides a sea surface target detection method, a detection platform and a system based on an optoelectronic pod.
In practical application, most of photoelectric pods are miniature or small structures, wherein multiple photoelectric detectors are used for collecting photon signals in different wave bands, for input photon signals in different wave bands, an image electric signal is generated through a photoelectric conversion circuit, then a plurality of single-channel image data are aggregated into one-channel high-speed image data through an image aggregation unit, and high-speed transmission of image data streams is realized through a single-channel high-speed link. The photoelectric detectors in the photoelectric pod can be flexibly configured according to task requirements, and the combination of detectors with different wave bands such as short wave infrared, visible light, long wave infrared and the like and lenses with different focal lengths such as wide-angle lenses, long-focus lenses and the like can be adopted. In the following embodiments, four photoelectric detectors are included in the photoelectric pod, which are respectively a combination of a visible light detector and a wide-angle lens, a combination of an infrared detector and a wide-angle lens, a combination of a visible light detector and a tele lens, and a combination of an infrared detector and a tele lens, so that the photoelectric pod can collect 4 paths of different sea surface image data at a time. For ease of description, in the following embodiments, these four paths of detectors are denoted as detector 1, detector 2, detector 3, and detector 4, respectively. It should be noted that the multiplex detector is only exemplary, and in other embodiments of the invention, a photoelectric pod with a different multiplex detector may be used to acquire sea surface image data.
The following are examples.
Example 1:
a sea surface target detection method based on a photoelectric pod, as shown in fig. 2, comprises:
continuously receiving sea surface image data transmitted by the photoelectric pod; and (3) executing the following steps (S1) - (S3) on the received multi-band sea surface image data after aggregation of each frame.
Step (S1) of the present embodiment includes: depolymerizing it into a plurality of single-pass image data; each single-path image data corresponds to image data of a wave band under one focal length collected by the optoelectronic pod; in this embodiment, after depolymerizing the received sea surface image data, four paths of image data, which are a visible light band wide-angle image, an infrared band wide-angle image, a visible light band long-focus image, and an infrared band long-focus image, are obtained. Optionally, in this embodiment, the visible light image size is 1920×1080, and the infrared image size is 640×512.
Step (S2) of the present embodiment includes: and respectively executing sea clutter self-adaptive suppression steps on each single-path image data to obtain corresponding denoised images.
Sea clutter noise with uniform brightness and simple texture on the sea surface, such as bright stripes, sea surface scale waves and the like, is a main component in sea surface clutter and can seriously interfere with detection of sea surface targets, so that the noise is firstly inhibited, and analysis and related experimental data show that top hat transformation can effectively inhibit the sea clutter noise; after sea clutter noise with uniform brightness and simple texture in a sea surface image is inhibited, noise with partial size close to that of a target but inconsistent shape still exists in the image, and the noise still interferes with a sea surface target detection result.
Based on the above analysis, as shown in fig. 2, in order to effectively suppress sea clutter, in this embodiment, the step of adaptive suppression of sea clutter includes:
(S21) performing top hat transformation on the image data to be processed to obtain a preliminary denoising image;
(S22) calculating gradients of the preliminary denoising image in different directions to obtain a plurality of gradient images;
(S23) taking the minimum pixel value of the plurality of gradient images at the same position as the pixel value of the denoised image at the position to obtain the denoised image corresponding to the image to be processed.
In the embodiment, the size of the target is further considered to be different in different scenes, so that the self-adaptive gradient convolution kernel is designed to cope with different scenes, and the suppression effect of sea clutter is further improved. Specifically, step (S22) includes:
carrying out connected domain marking on the preliminary denoising image, removing connected domains smaller than a preset threshold value, and calculating the average size C of the remaining connected domains;
Calculating a gradient convolution kernel size ks according to ks=c// M, and determining gradient convolution kernels in all directions according to the calculated gradient convolution kernel size;
Performing convolution operation on the preliminary denoising image by utilizing gradient convolution check under each direction to calculate gradients of the preliminary denoising image in different directions, so as to obtain a plurality of gradient images;
wherein "//" denotes integer division operations; in this embodiment, m=5, and experimental data indicates that, by setting the relationship between the gradient convolution kernel size and the average size of the connected domain, the effect of sea clutter can be suppressed to the maximum extent.
In this embodiment, the adaptive gradient calculation based on directionality aims to suppress noise with a size close to that of a target and a shape different from that of the target in an image, and after the noise is marked by a connected domain, each connected domain corresponds to one noise or a sea surface target, so that the undersize noise is removed according to a preset threshold value, and the average size of the connected domain can be ensured to be close to that of the sea surface target; alternatively, in the present embodiment, considering that the actual sea surface targets are mostly marine vessels, low-altitude unmanned aerial vehicles, and the like, the preset threshold is set to 10 based on the empirical sizes of these targets. In other embodiments, other values may be set according to the size of the actual detection target, so as to ensure that the communication domain that can be eliminated is significantly smaller than the target size.
The sea clutter adaptive suppression step is further explained below by taking a processing procedure of an actual image as an example. Fig. 3 shows original sea surface image data, in which the object in the box is the sea surface target to be detected, and the sea surface clutter self-adaptive suppression step according to the embodiment is used for processing the sea surface image data. After the sea surface image shown in fig. 3 is transformed by the top cap, the obtained image is shown in fig. 4, and comparing fig. 3 and fig. 4 can obviously show that bright stripes, sea surface scale waves and the like in the original image are effectively inhibited. After binarizing the image shown in fig. 4, labeling the connected domain, removing the connected domain smaller than 10, calculating according to ks=c// 5 to obtain gradient convolution kernels with the size of 7, and calculating gradient convolution kernels with the sizes of 8 directions of 0 °, 45 °,90 °, 135 °,180 °, 225 °, 270 ° and 315 ° respectively as shown in fig. 5, wherein gradient images with the corresponding directions can be obtained by calculating the gradient convolution kernels with the different directions. After gradient calculation is performed on the image shown in fig. 4, gradient diagrams in the directions of 0 ° and 90 ° are shown in fig. 6 and 7, respectively. And for the gradient images in 8 different directions, taking the minimum pixel value of each gradient image at the same pixel position as the pixel value of the corresponding position in the image, and obtaining the denoised image with sea clutter effectively suppressed.
In this embodiment, step (S3) includes: and respectively carrying out target detection on the denoised images under the same focal length, fusing detection results, and taking the fusion result as a sea surface target detection result under the corresponding focal length.
When the target detection is carried out on each path of image, a specific target detection algorithm can be selected according to the characteristics of the image under the corresponding wave band. In this embodiment, the visible light image size is 1920×1080, the infrared image size is 640×512, and as more information exists in the visible light image, YOLOv is selected as the visible light target detection algorithm, and the conventional multi-scale block contrast measurement algorithm (Multiscale Patch Contrast Measurement, MPCM) is selected to realize the detection of the infrared image target. In other embodiments of the present invention, the target detection algorithm for a specific image may be flexibly selected as other algorithms.
By fusing target detection results of images with different wave bands under the same focal length, different information carried by the images with different wave bands can be fully utilized, multi-wave band image cooperative processing is realized, and the detection precision of sea surface targets is further improved. Because images of different wave bands under the same focal length have the same view field, the detection results of the two images are matched with the detection frame, and in order to effectively realize fusion of the detection results of the images of different wave bands under the same focal length, the embodiment adopts a high-low dual-threshold fusion mode, specifically, for the image under a certain focal length, the target detection results are fused, and the method comprises the following steps:
calculating the overlapping degree (IOU) among detection frames obtained by target detection of each denoised image under the focal length, outputting the detection frames with the overlapping degree larger than a preset high threshold (for example, 0.7) as a trusted target, and rejecting the detection frames with the overlapping degree smaller than a preset low threshold (for example, 0.3) as false alarms;
Determining the track of the sea surface target according to the sea surface target detection result sequence of the sea surface image data before the current frame, and outputting the detection frame with the overlapping degree between a low threshold value and a high threshold value as a trusted target if the detection frame is positioned on the track of the sea surface target, otherwise, rejecting the detection frame as a false alarm.
The method comprises the steps of determining the track of a sea surface target according to a sea surface target detection result sequence of sea surface image data before a current frame, and completing the track based on the existing target tracking means.
Because the fields of view of the lenses with different focal lengths are different, the fields of view are larger, noise and interference objects are more, the target size is small, the detection difficulty is high, the fields of view are smaller, the noise and interference objects are smaller, the targets are clear, and the detection difficulty is small under the longer focal length, but sea surface targets are possibly located outside the fields of view, based on the fact, after the sea surface target detection result under the shorter focal length and the sea surface target detection result under the longer focal length are obtained at the same time, the sea surface target detection result under the longer focal length can be directly taken as a final sea surface target detection result; otherwise, if only the sea surface target detection result under one focal length is obtained, the sea surface target detection result is used as a final sea surface target detection result.
In order to further improve the detection precision of the sea surface target detection result, the invention further provides that the sea surface target detection result under the shorter focal length is regarded as a low confidence coefficient result, the offset information of the target center relative to the image center is calculated according to the low confidence coefficient result, and the information is used as guiding information for adjusting the pose of the longer lens by the photoelectric pod end, so that higher-quality target image data is obtained, and further, a more accurate long-focus image target detection result is obtained. Accordingly, the present embodiment further includes, after step (S3):
(S4) calculating the central position of the sea surface target according to the sea surface target detection result under the shorter focal length, and calculating the offset of the central position of the sea surface target relative to the center of the image;
And (S5) sending a command to the optoelectronic pod so that the optoelectronic pod adjusts the pose of the lens with a longer focal length according to the offset, and the central position of the sea surface target under the focal length is overlapped with the central position of the image.
In general, the photoelectric pod is used as a sea surface data acquisition end, so that multipath image data with different focal lengths and different wave bands can be acquired, and a more sufficient data basis is provided for accurately detecting a sea surface target; based on the characteristics of sea clutter, a sea clutter self-adaptive suppression step is provided, and sea clutter noise in sea surface image data can be effectively suppressed, so that the detection precision of a sea surface target is effectively improved. On the basis, the pose adjustment of the lens under a longer focal length in the photoelectric pod is guided by utilizing the target detection result under a shorter lens, so that higher-quality target image data can be obtained, and further, a more accurate long-focus image target detection result can be obtained.
Example 2:
a photoelectric pod-based sea surface target detection platform comprising:
a computer readable storage medium storing a computer program;
and a processor for reading a computer program stored in a computer-readable storage medium, and executing the sea surface target detection method based on the optoelectronic pod provided in the above embodiment 1.
The sea surface target detection platform based on the photoelectric pod provided by the embodiment can be any platform which can be used for carrying out data transmission with the photoelectric pod and has data processing capability, such as an airplane platform, a remote control platform and the like.
Example 3:
A sea surface target detection system, comprising:
the sea surface target detection platform based on the optoelectronic pod provided in the above embodiment 2;
an optoelectronic pod;
The image data stream is transmitted between the photoelectric pod and the sea surface target detection platform through a single channel link.
It will be readily appreciated by those skilled in the art that the foregoing description is merely a preferred embodiment of the invention and is not intended to limit the invention, but any modifications, equivalents, improvements or alternatives falling within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (7)
1. The sea surface target detection method based on the photoelectric pod is characterized by comprising the following steps of:
continuously receiving sea surface image data transmitted by the optoelectronic pod; for each received frame of aggregated multi-band sea surface image data, performing the steps of:
s1, depolymerizing the image data into a plurality of single-path image data; each single-path image data corresponds to image data of a wave band under a focal length acquired by the photoelectric pod;
S2, respectively executing sea clutter self-adaptive suppression steps on each single-path image data to obtain a corresponding denoised image; the sea clutter self-adaptive suppression step comprises the following steps:
S21, performing top hat transformation on image data to be processed to obtain a preliminary denoising image;
S22, calculating gradients of the preliminary denoising image in different directions to obtain a plurality of gradient images;
S23, taking the minimum pixel value of the plurality of gradient images at the same position as the pixel value of the denoised image at the position to obtain a denoised image corresponding to the image to be processed;
s3, respectively carrying out target detection on the denoised images under the same focal length, fusing detection results, and taking the fusion results as sea surface target detection results under the corresponding focal length;
The step S22 includes:
performing binarization on the preliminary denoising image, then performing connected domain marking, removing connected domains smaller than a preset threshold value, and calculating the average size C of the rest connected domains;
Calculating a gradient convolution kernel size ks according to ks=c// M, and determining gradient convolution kernels in all directions according to the calculated gradient convolution kernel size;
Performing convolution operation on the preliminary denoising image by utilizing gradient convolution check under each direction to calculate gradients of the binarized image in different directions, so as to obtain a plurality of gradient images;
wherein "//" denotes integer division operations; m is a preset positive integer.
2. The method for detecting a sea surface target based on a photoelectric pod according to claim 1, wherein m=5.
3. The optoelectronic pod-based sea surface target detection method of claim 1 or 2, further comprising, after step S3:
s4, calculating the central position of the sea surface target according to the sea surface target detection result under the shorter focal length, and calculating the offset of the central position of the sea surface target relative to the center of the image;
S5, sending an instruction to the photoelectric pod, so that the photoelectric pod adjusts the pose of a lens with a longer focal length according to the offset, and the central position of a sea surface target under the focal length is overlapped with the central position of an image.
4. The method for detecting a sea surface target based on the optoelectronic pod according to claim 1 or 2, wherein in the step S3, the target detection results under the same focal length are fused, including:
Calculating the overlapping degree between detection frames obtained by target detection of each denoised image under the focal length, outputting the detection frames with the overlapping degree larger than a preset high threshold as a trusted target, and rejecting the detection frames with the overlapping degree smaller than a preset low threshold as false alarms;
Determining the track of the sea surface target according to the sea surface target detection result sequence of the sea surface image data before the current frame, and outputting the detection frame with the overlapping degree between the low threshold value and the high threshold value as a trusted target if the detection frame is positioned on the track of the sea surface target, otherwise, rejecting the detection frame as a false alarm.
5. The method for detecting a sea surface target based on the optoelectronic pod according to claim 4, wherein in the step S3, if the denoised image is a visible light band image, the target detection is performed by YOLOv; and if the denoised image is an infrared band image, detecting by adopting a multi-scale block contrast measurement algorithm.
6. A photoelectric pod-based sea surface target detection platform, comprising:
a computer readable storage medium storing a computer program;
And a processor for reading a computer program stored in the computer readable storage medium and executing the sea surface target detection method based on the optoelectronic pod according to any one of claims 1 to 5.
7. A sea surface target detection system, comprising:
the optoelectronic pod based sea surface target detection platform of claim 6;
an optoelectronic pod;
and the image data stream is transmitted between the photoelectric pod and the sea surface target detection platform through a single-channel link.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410248872.3A CN117853932B (en) | 2024-03-05 | 2024-03-05 | Sea surface target detection method, detection platform and system based on photoelectric pod |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410248872.3A CN117853932B (en) | 2024-03-05 | 2024-03-05 | Sea surface target detection method, detection platform and system based on photoelectric pod |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117853932A CN117853932A (en) | 2024-04-09 |
| CN117853932B true CN117853932B (en) | 2024-05-14 |
Family
ID=90538545
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410248872.3A Active CN117853932B (en) | 2024-03-05 | 2024-03-05 | Sea surface target detection method, detection platform and system based on photoelectric pod |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117853932B (en) |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104504670A (en) * | 2014-12-11 | 2015-04-08 | 上海理工大学 | Multi-scale gradient domain image fusion algorithm |
| CN104751443A (en) * | 2014-12-12 | 2015-07-01 | 郑州轻工业学院 | Cotton fault detecting and identifying method based on multi-spectrum technology |
| CN108140239A (en) * | 2015-09-23 | 2018-06-08 | 皇家飞利浦有限公司 | For organizing the method and apparatus of identification |
| CN108364277A (en) * | 2017-12-20 | 2018-08-03 | 南昌航空大学 | A kind of infrared small target detection method of two-hand infrared image fusion |
| CN108764325A (en) * | 2018-05-23 | 2018-11-06 | 腾讯科技（深圳）有限公司 | Image-recognizing method, device, computer equipment and storage medium |
| CN109785314A (en) * | 2019-01-22 | 2019-05-21 | 中科院金华信息技术有限公司 | A kind of pck count detection system and method based on u-net network |
| CN111179334A (en) * | 2019-11-14 | 2020-05-19 | 青岛理工大学 | Sea surface small-area oil spilling area detection system and detection method based on multi-sensor fusion |
| CN111696107A (en) * | 2020-08-05 | 2020-09-22 | 南京知谱光电科技有限公司 | Molten pool contour image extraction method for realizing closed connected domain |
| CN112967305A (en) * | 2021-03-24 | 2021-06-15 | 南京莱斯电子设备有限公司 | Image cloud background detection method under complex sky scene |
| CN114972140A (en) * | 2022-04-11 | 2022-08-30 | 大连海事大学 | Sea surface infrared and visible light image fusion method based on target segmentation |
| CN115690402A (en) * | 2022-11-15 | 2023-02-03 | 吉林大学 | OCC image recognition decoding method based on gradient corner comprehensive detection |
| CN116248705A (en) * | 2022-11-29 | 2023-06-09 | 宜昌测试技术研究所 | Multichannel image transmission and processing system of miniature photoelectric pod |
| CN117274251A (en) * | 2023-11-20 | 2023-12-22 | 山东鲁抗医药集团赛特有限责任公司 | Tablet quality detection method in medicine production process based on image data |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1938249A2 (en) * | 2005-09-02 | 2008-07-02 | Blindsight, Inc. | A system and method for detecting text in real-world color images |
| JP2007140684A (en) * | 2005-11-15 | 2007-06-07 | Toshiba Corp | Image processor, method and program |
| US8891866B2 (en) * | 2011-08-31 | 2014-11-18 | Sony Corporation | Image processing apparatus, image processing method, and program |
| WO2020052668A1 (en) * | 2018-09-15 | 2020-03-19 | 北京市商汤科技开发有限公司 | Image processing method, electronic device, and storage medium |
-
2024
- 2024-03-05 CN CN202410248872.3A patent/CN117853932B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104504670A (en) * | 2014-12-11 | 2015-04-08 | 上海理工大学 | Multi-scale gradient domain image fusion algorithm |
| CN104751443A (en) * | 2014-12-12 | 2015-07-01 | 郑州轻工业学院 | Cotton fault detecting and identifying method based on multi-spectrum technology |
| CN108140239A (en) * | 2015-09-23 | 2018-06-08 | 皇家飞利浦有限公司 | For organizing the method and apparatus of identification |
| CN108364277A (en) * | 2017-12-20 | 2018-08-03 | 南昌航空大学 | A kind of infrared small target detection method of two-hand infrared image fusion |
| CN108764325A (en) * | 2018-05-23 | 2018-11-06 | 腾讯科技（深圳）有限公司 | Image-recognizing method, device, computer equipment and storage medium |
| CN109785314A (en) * | 2019-01-22 | 2019-05-21 | 中科院金华信息技术有限公司 | A kind of pck count detection system and method based on u-net network |
| CN111179334A (en) * | 2019-11-14 | 2020-05-19 | 青岛理工大学 | Sea surface small-area oil spilling area detection system and detection method based on multi-sensor fusion |
| CN111696107A (en) * | 2020-08-05 | 2020-09-22 | 南京知谱光电科技有限公司 | Molten pool contour image extraction method for realizing closed connected domain |
| CN112967305A (en) * | 2021-03-24 | 2021-06-15 | 南京莱斯电子设备有限公司 | Image cloud background detection method under complex sky scene |
| CN114972140A (en) * | 2022-04-11 | 2022-08-30 | 大连海事大学 | Sea surface infrared and visible light image fusion method based on target segmentation |
| CN115690402A (en) * | 2022-11-15 | 2023-02-03 | 吉林大学 | OCC image recognition decoding method based on gradient corner comprehensive detection |
| CN116248705A (en) * | 2022-11-29 | 2023-06-09 | 宜昌测试技术研究所 | Multichannel image transmission and processing system of miniature photoelectric pod |
| CN117274251A (en) * | 2023-11-20 | 2023-12-22 | 山东鲁抗医药集团赛特有限责任公司 | Tablet quality detection method in medicine production process based on image data |
Non-Patent Citations (2)
| Title |
|---|
| Gradient-based kernel selection technique for tumour detection and extraction of medical images using graph cut;Jyotsna Dogra， et al;IET Image Processing, 2020 - Wiley Online Library;20200101;第84-93页 * |
| 基于深度卷积神经网络的桥梁裂缝检测方法研究;胡文魁;广东:华南理工大学;20230715;第1-125页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117853932A (en) | 2024-04-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20210041555A1 (en) | System and method for camera radar fusion | |
| US12118781B2 (en) | Method and device for determining extraction model of green tide coverage ratio based on mixed pixels | |
| CN105973228A (en) | Single camera and RSSI (received signal strength indication) based indoor target positioning system and method | |
| WO2016112714A1 (en) | Assistant docking method and system for vessel | |
| CN106210484A (en) | Waters surveillance polynary associating sensing device and cognitive method thereof | |
| CN105730705B (en) | A kind of aircraft camera positioning system | |
| US20120249801A1 (en) | Image generation apparatus, image generation method and image generation program | |
| WO2011055772A1 (en) | Image target identification device, image target identification method, and image target identification program | |
| CN109657639B (en) | Situation awareness system and method based on panoramic vision | |
| CN110244314A (en) | One kind " low slow small " target acquisition identifying system and method | |
| KR101552095B1 (en) | Seaside surveillance system using radar and camera | |
| CN110515378A (en) | A kind of intelligent Target searching method applied to unmanned boat | |
| WO2021109457A1 (en) | Airport airplane automatic labeling method based on self-learning policy | |
| CN111626129A (en) | Ship target joint detection method based on satellite AIS and infrared camera | |
| EP4363877A1 (en) | Methods and systems for detecting vessels | |
| CN111323756A (en) | Deep learning-based marine radar target detection method and device | |
| CN113673385A (en) | Sea surface ship detection method based on infrared image | |
| CN112435249B (en) | Dynamic small target detection method based on circumferential scanning infrared search system | |
| CN111323757B (en) | Target detection method and device for marine radar | |
| CN116703986A (en) | Infrared and visible light-based transformation equipment temperature abnormality early warning method and system | |
| CN111460949B (en) | Real-time monitoring method and system for preventing external damage of power transmission line | |
| CN117853932B (en) | Sea surface target detection method, detection platform and system based on photoelectric pod | |
| CN115932834A (en) | Anti-unmanned aerial vehicle system target detection method based on multi-source heterogeneous data fusion | |
| CN111428573A (en) | Infrared weak and small target detection false alarm suppression method under complex background | |
| CN110134148A (en) | A kind of transmission line of electricity helicopter make an inspection tour in tracking along transmission line of electricity |
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 |