KR101134857B1 - Apparatus and method for detecting a navigation vehicle in day and night according to luminous state - Google Patents

Abstract

The present invention implements a vehicle detection function that can detect and track a driving vehicle according to the illumination condition using a camera that is projected around the vehicle. To this end, the present invention determines whether the illuminance state is a daytime state, a night state, an intermediate state such as dawn or rainy weather, the inside of a tunnel, and detects a vehicle using a vehicle detection method suitable for the illuminance state. In detail, when the illumination condition is in the daytime state, the vehicle is detected by using the feature points and the shadow of the vehicle.In the case of the night state, the vehicle is detected by the classification of the light source. Apply to detect the vehicle. In this way, since the vehicle detection method using the camera can be used in various illumination environments including day and night, it is possible to overcome the limitation that was applicable only in a specific environment and to develop a related system.

Vehicle Detection, Day, Night, Illumination

Description

TECHNICAL FIELD AND DEVICE FOR DETECTING DAY AND NIGHT LIGHTING CARS IN ACCORDING TO LIGHT CONDITIONS {APPARATUS AND METHOD FOR DETECTING A NAVIGATION VEHICLE IN DAY AND NIGHT ACCORDING TO LUMINOUS STATE}

1 is an internal block diagram of a vehicle detection apparatus according to an embodiment of the present invention;

2 is a control flowchart for performing vehicle detection according to an illuminance state according to an embodiment of the present invention;

3 is an exemplary view showing a process of calculating a roughness index according to an embodiment of the present invention;

4 is a view for explaining a vehicle detection process in a daytime state according to an embodiment of the present invention;

6 is an exemplary screen illustrating a process of setting a primary vehicle candidate region according to an embodiment of the present invention;

6 is an exemplary screen illustrating a process of setting a secondary vehicle candidate region according to an embodiment of the present invention;

7 is a view for explaining a vehicle detection process in a night state according to an embodiment of the present invention;

8 is an exemplary view for explaining a light source extraction process according to an embodiment of the present invention;

9 is a diagram illustrating a vehicle detection according to an illuminance state according to an embodiment of the present invention.

The present invention relates to a vehicle detection method and apparatus, and more particularly, to a method and apparatus for detecting a vehicle traveling day and night using a camera.

Frequently, a vehicle collides with a vehicle driving in front of a driver due to carelessness of a driver or a poor watch. That is, traffic accidents due to lane departure and collision with driving vehicles due to poor visibility and lack of driver's concentration in long distance driving, rainy weather, and night driving are constantly occurring. In order to prevent such a collision, a driver assistance system that can provide information such as collision warning, collision avoidance, and cruise control between vehicles by grasping the position and speed between vehicles through vehicle detection is provided.

In general, there is a method of detecting a vehicle in front of the vehicle using a radar, a camera, and the like, and a method of detecting an obstacle or a vehicle in a blind spot or rear parking is a method using an ultrasonic sensor.

Among them, in recent years, a vehicle detection method using a camera has been widely applied to a driver assistance system due to a decrease in the price of a camera module and an improvement in image processing performance. This method detects the vehicle using the prior knowledge of the vehicle as a feature point mainly during the day when the image can be obtained with high quality, such as shadow, edge, road color, symmetry and vehicle type model.

On the other hand, the vehicle is detected by using a low-light camera that operates at low light at night, and uses a bright area such as a vehicle body that is reflected and brightened by the headlights, taillights, brake lights, or other light sources of the vehicle.

However, the above methods are divided into daytime and nighttime conditions, respectively, and are applicable only to specific environments without considering the change in illumination. In addition, various illuminance such as rainy environment, tunnel interior, daytime, nighttime, dawn or sunset, etc. There is a limit to the application to the environment.

As described above, the vehicle detection method using a conventional camera is used only as a detection technology for each of the day and night conditions. Therefore, even if there are various environments such as a rainy environment, only one detection method was used. In addition, if the combination of the daytime vehicle detection method and the nighttime vehicle detection method is applied to the above environment, more errors are generated. Therefore, the technology and the technology for defining and detecting the feature points of the vehicle in the representative environment and verifying it with the vehicle There is a need for a technology that can be applied to various environments.

Accordingly, the present invention provides a method and apparatus for detecting day and night driving vehicles according to the illumination condition using a camera mounted on the vehicle.

According to an aspect of the present invention, there is provided a method of detecting day and night driving vehicles according to an illumination condition using a camera mounted on a vehicle, the method comprising: calculating an illumination index from an image input through the camera; Determining whether the calculated illuminance index is included in one of a range of a predetermined illuminance index divided into a pre-set day state, a night state, and an intermediate state; and as a result of the determination, the previous state in the image And detecting a vehicle candidate using a shadow area generated on the vehicle and the road in the image when the previous state is the daytime state.
In addition, the present invention is a method for detecting the day and night driving vehicles according to the illumination conditions using a camera mounted on the vehicle, the process of calculating the illumination index from the image input through the camera, and the calculated illuminance Determining whether the index is included in one of a range of illuminance indexes divided into a preset day state, a night state, and an intermediate state; and when the determination corresponds to the intermediate state, the vehicle corresponding to the previous state in the image. Detecting the vehicle, classifying the light sources in the image by the size of the light sources when the previous state is the night state, and detecting a vehicle candidate based on the classified light sources in the image. do.

In addition, the apparatus for detecting day and night driving vehicles according to the illumination conditions using a camera mounted on the magnetic vehicle according to the present invention, the image receiving unit for receiving the image through the camera, and receives the image from the image receiving unit The illuminance recognition unit configured to set a background part in the image, and calculate an illuminance index by measuring an average contrast of the set background part, and the illuminance index provided from the illuminance recognition unit is divided into a preset day state, night state, and intermediate state. An adjuster unit determining one of the ranges of the determined illuminance index ranges to determine one state, and detecting a vehicle corresponding to the determined state, and the image when the determined state corresponds to the intermediate state And a vehicle position detector for detecting a vehicle corresponding to the previous state at The.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. It should be noted that the same elements in the figures are represented by the same numerals wherever possible. In addition, in describing the present invention, when it is determined that a detailed description of a related known function or configuration may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted.

The present invention implements a vehicle detection function that can detect and track a driving vehicle according to the illumination condition using a camera that is projected around the vehicle. To this end, the present invention determines whether the illuminance state is a daytime state, a night state, an intermediate state such as dawn or rainy weather, the inside of a tunnel, and detects a vehicle using a vehicle detection method suitable for the illuminance state. In detail, when the illumination condition is in the daytime state, the vehicle is detected by using the feature points and the shadow of the vehicle.In the case of the night state, the vehicle is detected by the classification of the light source. Apply to detect the vehicle. In this way, since the vehicle detection method using the camera can be used in various illumination environments including day and night, it is possible to overcome the limitation that was applicable only in a specific environment and to develop a related system.

A function of the vehicle detection apparatus in which the above function is implemented will be described with reference to FIG. 1. 1 is an internal block diagram of a vehicle detection apparatus according to an embodiment of the present invention.

Referring to FIG. 1, the vehicle detecting apparatus according to the present invention includes an image receiving unit 101, an illuminance recognizing unit 102, an adjuster unit 103, and a vehicle position detecting unit 104.

First, the image receiving unit 101 refers to a camera mounted on a vehicle, and receives an image photographed by the camera and transmits it to the illuminance recognition unit 102. Then, the illumination recognition unit 102 separates the background portion from the image, measures the average contrast of the separated background portion, and converts the measured average contrast into an illuminance index to calculate it. Then, the illuminance recognition unit 102 transmits the calculated illuminance index to the adjuster unit 103 for adjusting the detection method. Accordingly, the adjuster unit 103 determines one state by determining whether the calculated illuminance index corresponds to one of daytime, nighttime, and intermediate state. That is, one state is determined by determining whether the calculated illuminance index is included in any one of a range of illuminance indexes divided into a preset day state, night state, and intermediate state.

If either state is determined, it is transmitted to the vehicle position detector 104. Accordingly, the vehicle position detector 104 selectively detects the vehicle using any one of a vehicle detection method corresponding to the determined state, for example, a daytime vehicle detection method or a nighttime vehicle detection method.

A process of selectively changing and detecting the vehicle detection method according to the illuminance state in the vehicle detection apparatus having the above-described configuration will be described with reference to FIGS. 2 and 3. 2 is a control flowchart for detecting a vehicle according to an illuminance state according to an embodiment of the present invention, and FIG. 3 is an exemplary diagram illustrating a process of calculating an illuminance index according to an embodiment of the present invention.

Referring to FIG. 2, the apparatus for detecting a vehicle acquires an image by using a camera in operation 200 and calculates a current illuminance index from the image obtained in operation 205. In this case, reference is made to FIG. 3 to describe in detail the process of calculating the roughness index. 3 (a) is a screen example of photographing a daytime running vehicle, and FIG. 3 (b) is a screen example of photographing a nighttime driving vehicle. First, in order to calculate the illuminance index, the image must be divided into a road part and a background part in which the vehicle travels. The motion is calculated for the entire image and the motion area is regarded as the area where the moving vehicle is located, and the area without motion is calculated. Can be separated into the background. At this time, there can be a variety of methods for separating the moving region and the non-moving region in the image, in the present invention, as indicated by reference numeral 10 of FIG. After setting a block having a size, the illuminance index is calculated assuming that it is a background portion. Similarly, the vehicle detecting apparatus can calculate an illuminance index for the block indicated by 20 in FIG. 3 (b).

The roughness index for this background portion is calculated through Equation 1 below.

는 배경 부분의 평균 명암값이며,

는 조도 지수이다. 상기 수학식 1을 통해서는 영상으로부터 배경 부분의 명암의 평균값

이 [0.255]까지의 자연수로 표현되는 조도 지수

를 얻을 수 있다. 이러한 조도 지수는 초당 10개의 영상을 획득하는 장치에서는 10개 영상 획득 시마다 갱신된다. 이와 같이 조도 지수는 영상에서 배경 부분으로 설정된 영역의 평균명암값으로부터 산출되는 것이며, 매 영상이 획득될 때마다 갱신되게 된다. In Equation 1,

Is the average contrast value of the background part,

Is the roughness index. Through Equation 1, the average value of the contrast of the background part from the image

Illuminance index expressed in natural numbers up to [0.255]

Can be obtained. The illuminance index is updated every time 10 images are acquired in a device that acquires 10 images per second. As such, the illuminance index is calculated from the average contrast value of the region set as the background portion of the image, and is updated each time the image is acquired.

When the illuminance index is calculated through Equation 1, the vehicle detecting apparatus determines an illuminance state corresponding to the illuminance index in step 210. That is, it is determined whether the calculated illuminance index is included in one of a range of illuminance indexes divided into a preset day state, night state, and intermediate state. In the present invention, the illuminance state is largely divided into a day state, a night state, and an intermediate state, and the middle state corresponds to a state where day and night feature points appear at the same time or a day or night is switched to dawn or sunset, inside a tunnel, or rainy weather. Thus, the state can be classified as shown in Table 1 according to the range of the illumination index for state transition.

[Table 1]

:[0, 70]Night state

: [0, 70]

:[71, 180]Intermediate

: [71, 180]

:[181, 255]Daytime

: [181, 255]

In the present invention, the illuminance state is divided according to the range of the illuminance index as shown in Table 1, but can be arbitrarily adjusted according to the performance of the camera, so that the middle region can include the region where day and night overlap, such as dawn or sunset. You can set the range of the roughness index.

Subsequently, the vehicle detecting apparatus determines an illuminance state corresponding to the illuminance index calculated based on Table 1, and determines whether the current illuminance state corresponds to a day state, a night state, or an intermediate state in step 215. If the vehicle is in the daytime state, the vehicle detecting apparatus proceeds to step 400 of FIG. 4. On the other hand, the vehicle detection apparatus proceeds to step 400 of the control flow of FIG. 4 when it is determined that the daytime state, the symbol 215 or 220 in Figures 2 and 4 to indicate that the step 400 is connected to 400 of FIG. B was used, and symbol A was used to indicate that step 200 of FIG. 2 is connected to step 420 of FIG. 4. Meanwhile, FIG. 4 illustrates a vehicle detection method in a daytime state, and the vehicle detection method is briefly described as follows.

First, the vehicle detection in the daytime state is extracted from the vehicle candidate using the vehicle feature points in the daytime and the shadow between the vehicle and the road, and then the vehicle candidate is determined using the left and right edge information, the symmetry, and the width of the vehicle. By verifying and tracking a region similar to the previous one in the next image around the region of interest in which the verified vehicle candidate is located. On the other hand, such a vehicle detection operation is made by the vehicle position detection unit, a more detailed description of Figure 4 will be described later.

In contrast, when the illuminance state is a night state, the vehicle detecting apparatus proceeds to step 700 of FIG. 7. Here, symbol C is used to indicate that step 215 or 220 of FIG. 2 is connected to step 700 of FIG. 7, and symbol 200 to indicate that step 200 of FIG. 2 is connected to step 730 of FIG. 7. A was used. A description of FIG. 7, which is a view for explaining a vehicle detection process in a night state, will be described later. Meanwhile, FIG. 7 illustrates a vehicle detection method in a night state. The vehicle detection method is schematically described as follows. Vehicle detection in the night state is performed by extracting a light source region illuminated by a headlight or a taillight from an image, and verifying and tracking a vehicle candidate using the extracted light source size.

On the other hand, when the illuminance state is an intermediate state, the vehicle detecting apparatus proceeds to step 220 to determine whether the previous state is a day state. In the case of the day state, the vehicle detection device proceeds to step 400 of FIG. do. In this case, the initial illuminance state may be set to a daytime state as a default. When the illuminance state is initially determined to be an intermediate state, the initial illuminance state may be regarded as a daytime state immediately, and step 220 may proceed to step 400 of FIG. 4. .

When the illuminance state is a daytime state through the above process, even if the vehicle detection method in the daytime state is applied, but the vehicle is being tracked by the vehicle detection method in the daytime state before the intermediate state, Vehicle detection method is applied. In this case, the vehicle detection method in the day state and the night state may be alternately applied to every image for the background portion excluding the vehicle. Similarly, even when the illuminance is in the midnight state, the vehicle detection method in the night state continues to be applied to the vehicle to be tracked once the vehicle is detected in the daytime state when the illuminance state of the background portion is changed to the daytime state. It can be applied. As described above, the vehicle detecting apparatus according to the present invention selectively tracks the vehicle in parallel with the vehicle detecting method in the day or night state according to the illumination condition.

Hereinafter, when the state divided by the illuminance index is a daytime state, a vehicle detection process in the daytime state according to an embodiment of the present invention will be described with reference to FIG. 4. In the following it will be described with reference to Figures 5 and 6 to facilitate understanding.

Referring to FIG. 4, the vehicle detection apparatus extracts a vehicle candidate from an image in operation 400, verifies a vehicle candidate extracted in operation 410, and detects a final vehicle in operation 420.

First, step 400 is described in detail, in order to effectively detect the shadow area generated on the vehicle and the road, the vehicle detecting apparatus removes the background part from the entire image 502 and moves the motion area, that is, in the region of interest 501. The normalized image 502 of FIG. 5 (a) and FIG. 5 (b) as shown in FIG. 5 (a) are enlarged by removing the background part and increasing the contrast difference between the pavement and the shadow area having relatively dark contrast through histogram equalization in the region of interest 501. To generate a binarized image 503. At this time, the road part is removed from the binarized image.

Then, the vehicle detecting apparatus scans the shadow area by scanning with a window 304 having a constant thickness in the direction of the arrow indicated by reference numeral 505 of FIG. 5C, that is, from the bottom to the top of the image. Search for. This window is used to search where the shadow is in the image, and the window size varies depending on the perspective. At this time, the vehicle detection apparatus also moves the window from the lower side to the upper direction in order to scan the entire region of interest 501 and also moves in the direction indicated by the arrow 506. By moving the window from left to right while moving from bottom to top as described above, the vehicle detecting apparatus scans the area of interest 501, and the vehicle detecting apparatus greatly reduces the contrast, but the region having the dark contrast of less than or equal to FIG. 5 (c). As described above, the shadow area 507 is detected and the shadow area 507 is set as the vehicle candidate area. In this way, the primary vehicle candidate region is set using the shadow region.

Then, in order to set a more precise vehicle candidate region, an operation of setting a vehicle candidate region secondary is performed. Referring to FIG. 6, in FIG. 6, after converting the vehicle candidate regions set as primary to actual coordinates on a plane, the secondary vehicle candidate regions are set by removing the crowded vehicle candidate regions outside the virtual lane on the plane. Show the process.

As shown in FIG. 6, the vehicle candidate regions are grouped into at least one or more points on a plane by removing perspective, and the vehicle candidate regions are divided into groups 601 and 602 by the virtual lane 603. Among the groups 601 and 602, the group 602 located outside the region of interest of the virtual lane 603 is removed, and the remaining group 601 is set as the secondary vehicle candidate region. As described above, FIG. 6 illustrates a result of processing the vehicle candidate regions extracted from the image by converting them into actual coordinates on a plane. Here, the coordinate transformation is performed through the following equation (2).

Equation (2) is a 3x3 matrix representing a transformation between image coordinates and actual coordinates, and the image coordinates ( r _i , c _i ) are three-dimensional actual coordinates ( x _i , y _i on a plane) by removing the perspective effect in the image. To).

In the vehicle candidate region extracted first through the above process, invalid vehicle candidate regions are again removed from the vehicle candidate regions extracted first through Equation 1 to generate the vehicle candidate region secondarily.

Hereinafter, a verification process for the secondary vehicle candidate region will be described. The verification of the vehicle candidate area is to determine whether the vehicle is actually located in the area where the vehicle is considered to be. For example, even if a secondary vehicle candidate region is obtained, an obstacle, etc., which is not a vehicle may be detected. Therefore, verification of the vehicle candidate region should be performed.

The secondary vehicle candidate region is verified by the symmetry, left and right vertical edge components, and the width of the vehicle, which are applied prior to the vehicle. Since the longitudinal position of the secondary vehicle candidate is set by the shadow region, it is relatively accurate, but since the lateral position is incorrect, it is reset to the region having excellent symmetry through the symmetry of the edge as shown in Equation 3 below.

In Equation 3, s is the number of edges having the same angle or opposite sign, and n is the total number of edges. As such, the vehicle detecting apparatus may determine which position the vehicle is located on the basis of the edge component obtained through Equation 3. For example, when the edge component is in the center, it can be determined that the position of the vehicle is also in the center. As described above, the vehicle candidate region is re-verified using the histogram of the left and right edges and the actual vehicle width information, thereby outputting the final vehicle position in the image. The vehicle detected through this process is stably detected in the continuous image through the vehicle tracking. The detected vehicle is set as a real-time vehicle model and updated every frame through reliability, and a conventional Lucas-Kanade method can be used for such tracking. If the tracking of the vehicle having high reliability fails, the vehicle may be restored by lowering the threshold set by the reliability.

Hereinafter, the vehicle detection process in the night state will be described with reference to FIG. 7. In order to help understand this, a process of extracting a light source will be described with reference to FIG. 8. 8 is an exemplary view for explaining a light source extraction process according to an embodiment of the present invention.

The vehicle detecting operation of FIG. 7 is performed by the vehicle position detecting unit, which is a component of the vehicle detecting apparatus, and hereinafter referred to as a vehicle detecting apparatus for convenience of description. Referring to FIG. 7, in operation 700, the vehicle detection apparatus bundles light sources such as headlights, taillights, and brakes in the night image. Since the light source has a high contrast of 250 or more in the image, the vehicle detection apparatus uses the same to binarize the image. In this case, when the light sources are bundled in an image represented by a light source, a small headlight, or the like, as shown in FIG. 8 (a), the light sources may be displayed as indicated by the green portion of FIG. 8 (b).

Then, the vehicle detecting apparatus proceeds to step 710 to determine the size of the light source, whether large or medium, small or small, such as whether it is a light source that spreads through a bundle of light sources or a small light source such as a taillight. Classify Each light source may have small, medium, or large sizes when the light source is displayed without distortion of the light source, such as a taillight or a brake, when the ambient light source is reflected by the vehicle, and when the headlight is greatly distorted. Accordingly, the vehicle detecting apparatus regards a light source corresponding to one vehicle as indicated by reference numeral 800 of FIG. 8C when the size of the light source is larger than the predetermined size through the light source classification. , As indicated by reference numeral 810 in FIG. 8 (c), the small taillights are collectively regarded as one vehicle. In this case, the threshold for classifying the size of each light source may be set by a ratio of the size of the actual light source to the size of the expected vehicle at the center of gravity of the corresponding light source. That is, light sources bound by the ratio of the size of the light source to the size of the vehicle at the center of gravity of the light source are classified into large, medium, and small. If the light sources are classified as they are, the vehicle detection device detects the vehicle candidate by removing the noise portion of the light source, for example, the distorted portion, and the light source classified into the small is combined with the ambient light source.

Based on the light sources classified as above, the vehicle detection apparatus extracts a vehicle candidate in operation 720. That is, the vehicle candidate region is generated by the size of the light source. In operation 730, the vehicle detecting apparatus tracks the light source by re-extracting the light source in the region of interest previously set in the next image for the stability of vehicle extraction.

 Meanwhile, reference will be made to FIG. 9 to describe a vehicle detection method in an intermediate state. 9 is a diagram illustrating a vehicle detection according to an illuminance state according to an embodiment of the present invention. 9A illustrates an example of a daytime running vehicle. Since the vehicle is in a daytime state at a tunnel entrance, the vehicle detecting apparatus applies a vehicle detection method in a daytime state to a vehicle. Subsequently, when the vehicle enters the tunnel as shown in FIG. 9 (b), the illuminance index may correspond to an intermediate state. Even when the illuminance index changes to an intermediate state, the vehicle detecting apparatus detects the vehicle in a daytime state with respect to the vehicle. Continue to apply to detect. In this case, the daytime vehicle detection method and the nighttime vehicle detection method are alternately applied to each image in the remaining regions except for the vehicle in FIG. 9 (b).

On the other hand, Figure 9 (c) is an exemplary view for explaining a situation where a new vehicle is detected instead of the vehicle being tracked, since the vehicle detection device is already tracking the vehicle in the tunnel using a night vehicle detection method, the new vehicle is detected. Even if the night vehicle detection method is used as it is to track a new vehicle.

As described above, according to the present invention, by providing a vehicle detection method that can be applied in various illumination environments including daytime and nighttime, the vehicle detection method using a camera overcomes the limitation that was applicable only in a specific environment to develop related systems. It becomes possible to plan.

Claims (14)

In the method for detecting the day and night driving vehicle according to the illumination condition using a camera mounted on his vehicle, Calculating an illuminance index from an image input through the camera; Determining whether the calculated illuminance index is included in one of a range of illuminance indexes divided into a preset day state, night state, and intermediate state; When the determination result corresponds to the intermediate state, detecting the vehicle corresponding to the previous state from the image; And detecting a vehicle candidate by using a shadow area generated on the vehicle and the road in the image when the previous state is the daytime state. The method of claim 1, And detecting the vehicle corresponding to the daytime state in the image, considering the daytime state if the previous state does not exist. . The method of claim 1, If the previous state is the daytime state, detecting the final vehicle by verifying the vehicle candidate by applying prior knowledge of the left and right edge components and the vehicle width to the detected vehicle candidate. Method for detecting the day and night driving vehicle according to the illumination conditions. The method of claim 1, Performing vehicle detection corresponding to the daytime state with respect to the vehicle candidate detected in the image when the previous state is the daytime state; The background portion of the image further includes the step of alternately performing the vehicle detection corresponding to the daytime state and the night state every image, the method for detecting the day and night driving vehicles according to the illumination conditions . In the method for detecting the day and night driving vehicle according to the illumination condition using a camera mounted on his vehicle, Calculating an illuminance index from an image input through the camera; Determining whether the calculated illuminance index is included in one of a range of illuminance indexes divided into a preset day state, night state, and intermediate state; When the determination result corresponds to the intermediate state, detecting the vehicle corresponding to the previous state from the image; When the previous state is the night state, classifying light sources in the image and classifying the light sources according to sizes of the light sources; And detecting a day and night driving vehicle according to an illumination condition, wherein the vehicle candidate is detected based on the classified light sources in the image. The method of claim 5, wherein the detecting of the vehicle candidate based on the classified light sources comprises: Classifying the light sources into one of large, medium, and small by the ratio of the size of the light source at the center of gravity of the light source to the size of the vehicle; Detecting a candidate for a vehicle by removing a noise portion of a light source classified as a large part of the image; The light source classified as the small in the image is a process of detecting the vehicle candidates by re-binding with the ambient light source according to the illumination conditions. The method according to claim 1 or 5, wherein the roughness index is It is calculated from the average contrast value of the area set as the background portion in the image, the method for detecting the day and night driving vehicles according to the illumination conditions, characterized in that each image is updated. In the device for detecting the day and night driving vehicle according to the illumination condition using a camera mounted on his vehicle, An image receiver for receiving an image through the camera; An illuminance recognition unit configured to receive the image from the image receiving unit, set a background portion in the image, and calculate an illuminance index by measuring an average contrast of the set background portion; An adjuster unit for determining one state by determining whether the illuminance index provided from the illuminance recognition unit is included in one of a range of illuminance indexes divided into a preset day state, night state, and intermediate state; Detects a vehicle corresponding to the determined state, detects a vehicle corresponding to a previous state in the image when the determined state corresponds to the intermediate state, and detects a vehicle in the image when the previous state is the day state And a vehicle position detector for detecting a candidate for a vehicle using a shadow area generated on the road and the road according to the illumination conditions. delete The method of claim 8, wherein the vehicle position detection unit, Apparatus for detecting day and night driving vehicles according to the illumination conditions characterized in that the vehicle candidates are detected by applying the detected vehicle candidates prior knowledge of left and right edge components and vehicle widths to verify the vehicle candidates. . The method of claim 8, wherein the vehicle position detection unit, In the absence of the previous state, the vehicle candidate is detected by using the shadow area generated on the vehicle and the road in the image by considering the daytime state, and the detected vehicle candidate is preset for the left and right edge components and the vehicle width of the vehicle. And detecting the final vehicle by applying knowledge to detect the final vehicle according to the illumination conditions. The method of claim 8, wherein the vehicle position detection unit When the determined state is the daytime state, the vehicle candidate is detected by using the shadow area generated on the vehicle and the road in the image, and the detected vehicle candidate is applied with prior knowledge about left and right edge components and the vehicle width of the vehicle candidate. And detecting the final vehicle by verifying the vehicle candidate, according to the illumination conditions. The method of claim 8, wherein the vehicle position detection unit When the determined state is the night state, the light sources are grouped according to the size of the light sources by grouping the light sources in the image, and the vehicle candidates are detected based on the classified light sources in the image. The device according to the detection. The method of claim 13, wherein the vehicle position detection unit The light sources are classified into one of large, medium, and small sizes according to the ratio of the size of the light source at the center of gravity of the light source to the size of the vehicle. Vehicle candidates are detected by removing a noise portion of a light source classified as a large and a light source classified as a small in the image is again grouped with an ambient light source to detect a vehicle candidate. To detect the light according to the illumination condition.

Images