JP7155870B2 - VEHICLE DETECTION APPARATUS AND METHOD AND ELECTRONICS - Google Patents

Description

TECHNICAL FIELD The present invention relates to the field of information technology, and more particularly to a vehicle detection apparatus and method and electronic equipment.

As city traffic conditions become more complex, the need for traffic condition monitoring and control increases. Also, detection of moving vehicles is one of the functions often used in video surveillance. At present, vehicle detection methods typically use classifiers or the like to detect and count vehicles in surveillance images for scenes with relatively good lighting conditions, such as daytime. However, in scenes with relatively poor lighting conditions, such as at night, it is difficult to effectively detect vehicles by conventional vehicle detection methods.

Conventionally, the general follow-up of vehicle detection for scenes with relatively poor lighting conditions is as follows: acquire a binary image based on the grayscale image of the surveillance video image; Region extraction is performed, then pairing is performed for each high brightness region, and vehicle position is determined based on the pairing of high brightness region pairing success.

The inventors of the present invention have discovered the following. That is, when a vehicle is detected using the conventional method as described above, generally, detection can be performed only for a high-brightness area where the headlights of the vehicle are located. Moreover, since a grayscale image is used, false detection is likely to occur. In addition, blinking of the vehicle's direction indicator lights (winkers) may also affect the detection results.

Embodiments of the present invention provide a vehicle detection apparatus, method and electronics for detecting directly on the color image and performing two different pairing processes to detect the headlights and taillights of the vehicle. can be detected and paired at the same time, and can avoid affecting the detection result of the vehicle's turn signal lights, which greatly reduces the possibility of detection omission and false detection, and the vehicle The comprehensiveness and accuracy of detection results can be effectively improved.

According to a first aspect of an embodiment of the present invention, there is provided a vehicle detection device, said device comprising: a first determining unit for determining high brightness regions in an input color image; an expansion unit for expanding the high-brightness region based on the color of the pixel; a first pairing unit for first pairing the high-brightness region having a predetermined color after the expansion, wherein, selecting for each high-brightness region based on the first set of parameters, and pairing each high-brightness region after selection based on the size difference between any two high-brightness regions to obtain the first set of high-brightness region pairs; , based on the second set of parameters to select the high-brightness region of the first pairing failure, and for each high-brightness region after selection based on the shape matching degree between any two high-brightness regions obtaining a second set of high-brightness region pairs by performing pairing; and based on the first set of high-brightness region pairs and the second set of high-brightness region pairs, input a second determining unit for determining the position of the vehicle in the color image;

According to a second aspect of an embodiment of the invention, there is provided an electronic device comprising the vehicle detection device according to the first aspect of an embodiment of the invention.

According to a third aspect of an embodiment of the present invention, there is provided a vehicle detection method, the method determining high intensity regions in an input color image; based on the color of pixels surrounding each high intensity region; , expanding the high-brightness regions; performing a first pairing for the high-brightness regions having a predetermined color after the expansion, wherein each high-brightness region is selected according to a first set of parameters; and performing pairing for each selected high-brightness region based on the size difference between any two high-brightness regions to obtain a first set of high-brightness region pairs; The pairing-failed high-brightness region is paired for the second time, wherein the first pairing-failed high-brightness region is selected according to the second set of parameters, and any two high-brightness regions are selected. obtaining a second set of high-brightness region pairs by performing pairing for each selected high-brightness region based on the degree of shape matching between the brightness regions; and the first set of high-brightness regions. determining the position of the vehicle in the input color image based on the pairs and the second set of high intensity region pairs;

Beneficial effects of the present invention are as follows. That is, by performing direct detection on the color image and performing two different pairing processes, it is possible to perform simultaneous detection and pairing on the headlights and taillights of the vehicle, and also on the vehicle. This can avoid the influence on the detection result of the direction indicator light, thereby greatly reducing the possibility of detection omission and false detection, and effectively improving the completeness and accuracy of the vehicle detection result.

FIG. 2 is a diagram showing a vehicle detection method in Example 1 of the present invention; FIG. 4 is a diagram showing another vehicle detection method in Embodiment 1 of the present invention; FIG. 10 is a diagram showing a vehicle detection device in Example 2 of the present invention; FIG. 10 is a diagram showing another vehicle detection device in Example 2 of the present invention; FIG. 10 is a diagram showing an electronic device according to Example 3 of the present invention; FIG. 10 is a diagram showing the system configuration of an electronic device according to Example 3 of the present invention;

Preferred embodiments for carrying out the present invention will now be described in detail with reference to the accompanying drawings.

This embodiment provides a vehicle detection method. FIG. 1 is a diagram showing a vehicle detection method according to Example 1 of the present invention. As shown in FIG. 1, the method includes the following steps.

Step 101: Determine a high-brightness region (a high-brightness region is, for example, a region with a predetermined brightness or higher) in the input color image;
Step 102: Expand each high-brightness region based on the color of pixels around the high-brightness region;
Step 103: perform a first pairing for the high-brightness regions with a predetermined color after the extension, wherein each high-brightness region is selected according to the first set of parameters, and any two obtaining a first pair of high-brightness regions by pairing each high-brightness region after selection based on the size difference between the high-brightness regions;
Step 104: Perform a second pairing for the high-brightness region that failed the first pairing, and select the high-brightness region that failed the first pairing based on the second set of parameters. , obtaining a second set of high-brightness region pairs by performing pairing for each selected high-brightness region according to the degree of shape matching between any two high-brightness regions;
Step 105: Determine the position of the vehicle in the input color image according to the first set of high-brightness region pairs and the second set of high-brightness region pairs.

As can be seen from this example, it is possible to detect and pair simultaneously to the vehicle headlights and taillights by performing direct detection on the color image and performing two different pairing processes. and avoid affecting the detection result of the vehicle's turn signal lights, which greatly reduces the possibility of detection omission and false detection, and effectively improves the completeness and accuracy of the vehicle detection result. can be improved.

In this embodiment, the input color image may be a frame of surveillance video, which may be captured by a camera mounted above the area that needs to be monitored. .

At step 101, a high brightness region in an input color image is determined, for example, a grayscale image of the input color image is obtained, a binary image is extracted from the grayscale image, and a binary image is extracted from the grayscale image. Based on the image, high brightness regions in the input color image may be determined.

In this embodiment, all high-brightness regions in a partial region of the input color image may be determined, for example, all regions of interest (Region of Interest, ROI) in the input color image A high brightness area may be determined. The ROI area is, for example, an area where roads are located.

In this embodiment, the binarization threshold can be adaptively adjusted.

At step 102, the high intensity regions may be expanded based on pixels of a predetermined color around each high intensity region, e.g. The expansion may be performed until there are no more pixels of . That is, each high-brightness region after expansion includes pixels of a predetermined color around each high-brightness region before expansion.

In this embodiment, the predetermined color may be the color when the tail lights are turned on, such as red, yellow, or orange. In this way, it is possible to include all areas where differently shaped taillights are located in the high brightness area.

In this embodiment, for example, the color of a pixel can be determined by converting a color image to HSV space and comparing the hsv scalar with the HSV range of each color according to experience.

In step 103, after the expansion, the high-intensity regions having a predetermined color are paired for the first time. A first pair of high-brightness regions is obtained by pairing each high-brightness region after selection based on the size difference between the two high-brightness regions.

In this embodiment, the predetermined color may be the color when the tail lights are turned on, such as red, yellow, or orange. In this way, it is possible to perform the first pairing for the high-brightness region having the predetermined color after the expansion, that is, perform the first pairing for the region where the taillight may exist. can.

In this embodiment, the first set of parameters may be determined according to the actual situation, for example, the first set of parameters are the vertical coordinate value of the bright region, the aspect ratio of the bounding rectangle, At least one of area difference, color difference, and degree of shape matching may be included.

When performing the first pairing, each high-brightness region is selected according to the first set of parameters, and the specific selection conditions may be set according to the actual situation. Among them, when the selection is made based on the degree of shape matching, the selection conditions may be set relatively loosely.

In this embodiment, the degree of shape matching can be weighed, for example, by a shape matching value, which can be obtained by conventional algorithms. For example, the larger the shape matching value, the greater the difference in shape.

After making a selection for each high intensity region based on the first set of parameters, pair each selected high intensity region based on the size difference between any two high intensity regions. , the first set of high-brightness region pairs can be configured from all high-brightness region pairs for which pairing is successful.

For example, for seven bright regions A to G after making a selection for each bright region based on the first set of parameters, calculate the size difference between any two bright regions . The size difference between A and D is the smallest of the size differences between A and all regions B to G, and the size difference between D and A is greater than D and A to C and E to G A and D are said to be successfully paired (also called pairing success) if the size difference with all regions of B is the smallest; The smallest difference in size from all regions to G, but the difference in size between E and B is not the smallest difference in size between E and all regions A to D and F to G In this case, it can be said that pairing between B and E has failed (also called pairing failure).

In this embodiment, the size difference may be the sum of the size differences in each dimension of the high-brightness regions, for example, the size difference is the width difference and the height of the two high-brightness regions. It may be a weighted sum of differences.

In step 104, perform a second pairing for the high-brightness regions that failed the first pairing, and select the high-brightness regions that failed the first pairing based on the second set of parameters. and pairing each selected high-brightness region based on the degree of shape matching between any two high-brightness regions to obtain a second set of high-brightness region pairs.

In this embodiment, the second set of parameters may be determined according to the actual situation, for example, the second set of parameters are the vertical coordinate value of the bright region, the aspect ratio of the bounding rectangle , and area difference.

After selecting each high-brightness region based on the second set of parameters, pairing each high-brightness region after selection based on the degree of shape matching between any two high-brightness regions , the second set of high-brightness region pairs can be configured from all the high-brightness region pairs for which pairing is successful.

For example, similarly, for seven high-intensity regions A to G after making a selection for each high-intensity region based on the second set of parameters, the shape matching value between any two high-intensity regions to calculate The shape matching value of A and D is the smallest of the shape matching values of A and all regions B to G, and the shape matching value of D and A is the same as D and A to C and E to G A and D are said to be paired successfully (also called pairing success) if the shape matching value with all regions of is the smallest; The shape matching value between E and B is the smallest among the shape matching values with all regions to G, but the shape matching value between E and B is not the minimum among the shape matching values between E and all regions A to D and F to G In this case, it can be said that pairing between B and E has failed (also called pairing failure).

In this example, the high-brightness region where the taillights in the color image are located has a predetermined color, so it receives the first pairing and the second pairing, but the headlights in the color image For the high-intensity region where is located, it is generally a white region and does not have a predetermined color, so it is excluded during the first pairing and goes directly into the second pairing.

At step 105, the position of the vehicle in the input color image is determined based on the first set of high brightness region pairs and the second set of high brightness region pairs.

For example, based on all bright region pairs in the first set of bright region pairs and the second set of bright region pairs, a normal (predetermined) ratio of car lights to car width is can be used to determine the position of the vehicle in the input color image.

FIG. 2 is a diagram showing another vehicle detection method according to the first embodiment of the present invention. As shown in FIG. 2, the method includes the following steps.

Step 201: Determine a bright region in the input color image;
Step 202: Expand each high-brightness region based on the color of pixels around the high-brightness region;
Step 203: Make a first rational selection for each high-brightness region after expansion, wherein for each high-brightness region after expansion, the area of the high-brightness region and/or the area of the high-brightness region making a selection based on convexity;
Step 204: performing a first pairing for the high-brightness regions with a predetermined color after the first rational selection after expansion to obtain a first set of high-brightness region pairs;
Step 205: perform a second pairing on the high-brightness regions that failed the first pairing to obtain a second set of high-brightness region pairs;
Step 206: Make a second rational selection for the second set of high-brightness region pairs, wherein the size of each high-brightness region pair in the second set of high-brightness region pairs, the color difference of the center of mass , the intensity difference of the centroid, and the intensity histogram of the bounding rectangle, and merging based on the degree of overlap of each pair of high intensity regions, yielding 2 obtaining a second set of high intensity region pairs after the rational selection for the second time;
Step 207: Determine whether a predetermined condition is satisfied, such as whether the first pairing, the second pairing, and the second reasonable selection are repeated a predetermined number of times, or , whether or not there are newly increased high-intensity region pairs with successful pairing. If the determination result is "yes", go to step 208; if the determination result is "no", return to step 204;
Step 208: Add a virtual high-brightness region corresponding to the single high-brightness region of pairing failure, and divide each high-brightness region pair consisting of the single high-brightness region and its corresponding virtual high-brightness region into a third set. As bright region pairs;
Step 209: Based on the first set of high-brightness region pairs, the second set of high-brightness region pairs after a second reasonable selection, and the third set of high-brightness region pairs, an input color Determine the position of the vehicle in the image.

In this embodiment, steps 201, 202, 204, 205 and 209 can be performed in the same manner as steps 101-105 above, and the detailed description thereof is omitted here.

In step 203, a first rational selection is made for each high-brightness region after expansion, wherein for each high-brightness region after expansion, the area of the high-brightness region and/or the convexity of the high-brightness region Make a selection based on (convexity). After making the first reasonable selection, the bright regions that obviously do not belong to headlights and taillights can be removed.

In this embodiment, the convexity can represent the degree of regularity of a shape, for example, the convexity of a region is defined by the area of the region and the minimum envelope enveloping the region. ) is the ratio of the circumference of

In step 206, a second rational selection is made for a second set of high-intensity region pairs, wherein the size of each high-intensity region pair in the second set of high-intensity region pairs, the center of mass point A second reasonable A second set of high-luminance region pairs after the selection is obtained. In this manner, erroneous detection due to road surface reflection and the like, and duplicate detection of a vehicle when the vehicle has multiple pairs of lights can be eliminated, thereby further improving the accuracy of detection results.

In this embodiment, by repeating steps 204 to 206, the possibility of detection omissions can be further reduced, and the completeness and accuracy of detection results can be further improved.

In this embodiment, the number of overlapping (repeated) times may be a predetermined number of times, for example, it may be repeated once. Alternatively, steps 204-206 may be repeated until there are no newly increased pairs of high intensity regions with successful pairing.

At step 208, add a virtual high-brightness region corresponding to the single high-brightness region of pairing failure, and divide each high-brightness region pair consisting of the single high-brightness region and its corresponding virtual high-brightness region into a third set. In this way, factors such as ROI region limitations, occlusions, and accidental mismatches can lead to unsuccessful matching, but the high-brightness region, which is actually the car light region, is added to the detection result. This can further reduce the possibility of detection failures and further improve the completeness and accuracy of detection results.

As can be seen from this example, it is possible to detect and pair simultaneously to the vehicle headlights and taillights by performing direct detection on the color image and performing two different pairing processes. and avoid affecting the detection result of the vehicle's turn signal lights, which greatly reduces the possibility of detection omission and false detection, and effectively improves the completeness and accuracy of the vehicle detection result. can be improved.

This embodiment provides a vehicle detection device, which corresponds to the vehicle detection method described in Embodiment 1, so for the specific implementation of the vehicle detection device, refer to the implementation of the vehicle detection method in Embodiment 1. possible, and redundant explanations will be omitted.

FIG. 3 is a diagram showing a vehicle detection device according to Example 2 of the present invention. As shown in FIG. 3, the vehicle detection device 300 includes the following.

A first determining unit 301: determining a high-brightness region in the input color image;
Expansion unit 302: expanding each high-brightness region based on the color of pixels around the high-brightness region;
a first pairing unit 303: performing a first pairing for the high-brightness regions with a predetermined color after the expansion, wherein each high-brightness region is selected according to the first set of parameters; obtaining a first set of high-brightness region pairs by pairing each high-brightness region after selection based on the size difference between any two high-brightness regions;
A second pairing unit 304: perform a second pairing for the high-brightness region with the first pairing failure, wherein the high-brightness region with the first pairing failure is paired according to the second set of parameters; and obtaining a second set of high-brightness region pairs by performing pairing for each selected high-brightness region according to the degree of shape matching between any two high-brightness regions;
A second determining unit 305: determining the position of the vehicle in the input color image according to the first set of high-brightness region pairs and the second set of high-brightness region pairs.

FIG. 4 is a diagram showing another vehicle detection device according to Example 2 of the present invention. As shown in FIG. 4, the vehicle detection device 400 includes the following.

A first determining unit 401: determining a high-brightness region in the input color image;
Expansion unit 402: expanding each high-brightness region based on the color of pixels around the high-brightness region;
A first selection unit 403: Before the first pairing, make a first rational selection for each extended high-brightness region, among which, for each extended high-brightness region, select the making a selection based on the area of the bright regions and/or the convexity of the bright regions;
A first pairing unit 404: perform a first pairing for the high-brightness region with a predetermined color after the first reasonable selection after expansion, wherein each high-brightness region is paired according to the first set of parameters; A first set of high-intensity region pairs is obtained by making a selection on the luminance region and pairing for each selected high-intensity region based on the size difference between any two high-intensity regions. to obtain;
A second pairing unit 405: perform a second pairing for the high-brightness region with the first pairing failure, wherein the high-brightness region with the first pairing failure is paired according to the second set of parameters; and obtaining a second set of high-brightness region pairs by performing pairing for each selected high-brightness region according to the degree of shape matching between any two high-brightness regions;
A second selection unit 406: after the second pairing, make a second rational selection for the second set of high-brightness region pairs, among which the second set of high-brightness region pairs The selection is based on at least one of the size of each high-brightness region pair, the color difference of the center of mass, the brightness difference of the center of mass, and the brightness histogram of the circumscribing rectangle, and the degree of overlap of each high-brightness region pair is obtaining a second set of high intensity region pairs after the second rational selection by merging based on;
Judgment unit 407: Judging whether the first pairing, the second pairing, and the second reasonable selection have been repeated a predetermined number of times, or whether there is a newly increased pairing successful high-brightness region pair. death;
Adding unit 408: adding a virtual high intensity region corresponding to the single high intensity region of pairing failure, and substituting each high intensity region pair consisting of the single high intensity region and its corresponding virtual high intensity region into a third set of high intensity regions; as luminance region pairs;
A second determining unit 409: based on the first set of high-brightness region pairs, the second set of high-brightness region pairs after the second rational selection, and the third set of high-brightness region pairs, input determining the position of the vehicle in the color image obtained.

As can be seen from this example, it is possible to detect and pair simultaneously to the vehicle headlights and taillights by performing direct detection on the color image and performing two different pairing processes. and avoid affecting the detection result of the vehicle's turn signal lights, which greatly reduces the possibility of detection omission and false detection, and effectively improves the completeness and accuracy of the vehicle detection result. can be improved.

This embodiment provides an electronic device. FIG. 5 is a diagram showing an electronic device according to Example 3 of the present invention. As shown in FIG. 5, the electronic device 500 includes a vehicle detection device 501, and the structure and function of the vehicle detection device 501 are the same as those described in the second embodiment, so detailed description thereof is omitted here. do.

FIG. 6 is a diagram showing the system configuration of an electronic device according to Example 3 of the present invention. As shown in FIG. 6, the electronic device 600 may include a central processing unit 601 and a memory 602 , the memory 602 being connected to the central processing unit 601 . It should be noted that the diagram is only an example, and that the structure can be supplemented or substituted with other types of structures to implement telecommunications functions or other functions.

The electronic device 600 may further include an input unit 603, a display 604, and a power supply 605, as shown in FIG.

In one implementation, the functionality of the vehicle detection device described in Example 2 can be integrated into the central processing unit 601 . Wherein, the central processing unit 601 may be configured as follows: determine the high-brightness regions in the input color image; expanding the luminance region; after the expansion, performing a first pairing for the high luminance region with a predetermined color, wherein each high luminance region is selected according to the first set of parameters, and optionally obtain a first set of high-brightness region pairs by pairing each high-brightness region after selection based on the size difference between the two high-brightness regions; perform the second pairing for the high-brightness regions, among which, based on the second set of parameters, select the high-brightness regions that failed the first pairing, and select any two high-brightness regions A second pair of high-brightness regions is obtained by pairing each high-brightness region after selection based on the degree of shape matching between the first pair of high-brightness regions and the Determine the position of the vehicle in the input color image based on the second set of high intensity region pairs.

For example, expanding the high-intensity region based on the color of pixels surrounding each high-intensity region includes expanding the high-intensity region to include pixels of a predetermined color around each high-intensity region.

For example, the central processing unit 601 may be further configured to: perform a first rational selection for each enhanced high intensity region before performing the first pairing; wherein, for each high-brightness region after extension, making a selection based on the area of the high-brightness region and/or the convexity of the high-brightness region; and, a high-brightness region having a predetermined color after extension. , and performing the first pairing for the high-brightness region having a predetermined color after extension, after the first rational selection after extension, It includes performing a first pairing on a high intensity region having a predetermined color.

For example, the central processing unit 601 may be further configured to: after performing the second pairing, a second reasonable making a selection, among which at least one of a size of each high-brightness region pair in the second set of high-brightness region pairs; and merging based on the degree of overlap of each high-brightness region pair to obtain a second set of high-brightness region pairs after the second reasonable selection, and the first determining the position of the vehicle in the input color image based on a set of high-brightness region pairs and the second set of high-brightness region pairs; Determining the position of the vehicle in the input color image based on a set of high-brightness region pairs comprises: the first set of high-brightness region pairs and the second set after a second rational selection; determining the position of the vehicle in the input color image based on the high intensity region pairs of .

For example, the central processing unit 601 may be further configured to repeat the first pairing, the second pairing, and the second rational selection a predetermined number of times; Alternatively, the first pairing, the second pairing, and the second rational selection are repeated until there is no newly increased pairing-successful high-brightness region pair.

For example, the central processing unit 601 may be further configured to: add a virtual bright region corresponding to a single bright region of pairing failure; Each high-brightness region pair consisting of the corresponding virtual high-brightness regions is a third set of high-brightness region pairs, and based on the first set of high-brightness region pairs and the second set of high-brightness region pairs, determining a position of a vehicle in the input color image; and determining a position of the vehicle in the input color image based on the first set of high-brightness area pairs and the second set of high-brightness area pairs. determining a position based on the first set of high intensity region pairs, the second set of high intensity region pairs after the second rational selection, and the third set of high intensity region pairs; determining the position of the vehicle in the input color image;

In another implementation, the vehicle detection device described in Example 2 may be configured separately from the central processing unit 601, for example, the vehicle detection device may be configured as a chip connected to the central processing unit 601. However, the functions of the vehicle detection device may be realized under the control of the central processing unit 601 .

In this embodiment, the electronic device 600 need not necessarily include all the components shown in FIG.

As shown in FIG. 6, the central processing unit 601, sometimes referred to as a controller or operational controller, may include a microprocessor or other processing and/or logic device, and the central processing unit 601 , can receive input to control the operation of each component of the electronic device 600 .

The storage 602 may be, for example, one or more of a buffer, flash memory, HDD, removable media, volatile storage, non-volatile storage, or other suitable device. The central processing unit 601 can implement information storage and processing by executing a program stored in the storage unit 602 . Since the functions of other parts are the same as those of the conventional one, detailed description thereof will be omitted here. Also, each component of the electronic device 600 may be implemented by dedicated hardware, firmware, software, or a combination thereof, all of which are within the scope of the present invention.

As can be seen from this example, it is possible to detect and pair simultaneously to the vehicle headlights and taillights by performing direct detection on the color image and performing two different pairing processes. and avoid affecting the detection result of the vehicle's turn signal lights, which greatly reduces the possibility of detection omission and false detection, and effectively improves the completeness and accuracy of the vehicle detection result. can be improved.

An embodiment of the present invention further provides a computer readable program, wherein when the program is executed in the vehicle detection device or the electronic device, the program causes the computer to read the code of the embodiment in the vehicle detection device or the electronic device. 1 can be executed.

An embodiment of the present invention further provides a storage medium storing a computer readable program, wherein the computer readable program instructs a computer to perform the vehicle detection method according to embodiment 1 in a vehicle detection device or an electronic device. can be executed.

Also, the apparatus, method, etc. according to the embodiments of the present invention may be implemented by software, hardware, or a combination of hardware and software. The invention also relates to such a computer readable program, i.e. said program, when executed by a logic component, is capable of causing said logic component to implement a device or component as described above, or The logic component may implement the above method or steps thereof. Furthermore, the present invention also relates to a storage medium for storing the above program, such as a hard disk, magnetic disk, optical disk, DVD, fresh memory, and the like.

In addition, the following supplementary remarks are disclosed with respect to the above examples and the like.

(Appendix 1)
A vehicle detection device,
a first determining unit for determining a high brightness region in an input color image;
an expansion unit for expanding the high brightness region based on the color of the pixels surrounding each high brightness region;
A first pairing unit for performing a first pairing for the extended high-brightness regions having a predetermined color, wherein each high-brightness region is selected based on a first set of parameters. , a first pairing unit for obtaining a first pair of high-brightness regions by pairing each high-brightness region after selection based on the size difference between any two high-brightness regions; ;
A second pairing unit for performing a second pairing to the first pairing-failed high-brightness region, the second pairing unit for the first pairing-failed high-brightness region based on a second set of parameters. , and pairing each selected high-brightness region based on the degree of shape matching between any two high-brightness regions to obtain a second set of high-brightness region pairs. a second pairing unit; and a second determination unit for determining the position of the vehicle in the input color image based on the first set of bright region pairs and the second set of bright region pairs. apparatus, including

(Appendix 2)
A device according to Appendix 1, wherein
The apparatus of claim 1, wherein the expanding unit expands the high-brightness regions to include pixels of a predetermined color around each high-brightness region.

(Appendix 3)
1. The device of clause 1, further comprising:
A first selection unit for making a first rational selection for each extended high-brightness region before performing the first pairing, wherein for each extended high-brightness region: a first selection unit that makes a selection based on the area of the high-brightness region and/or the convexity of the high-brightness region;
The apparatus, wherein the first pairing unit performs the first pairing for a high-brightness region having a predetermined color after the first rational selection after expansion.

(Appendix 4)
3. The apparatus of Supplementary Note 3, further comprising:
A second selection unit for making a second rational selection for the second set of high-brightness region pairs after performing a second pairing, wherein the second set of high-brightness region pairs is selection based on at least one of the size of each high-brightness region pair in a second selection unit for obtaining a second set of high intensity region pairs after the second rational selection by merging based on
The second determination unit determines the position of the vehicle in the input color image based on the first set of high-brightness region pairs and the second set of high-brightness region pairs after a second reasonable selection. Determine the device.

(Appendix 5)
The device of Appendix 4, wherein
The first pairing unit, the second pairing unit, and the second selection unit repeat the first pairing, the second pairing, and the second rational selection a predetermined number of times, Alternatively, the apparatus repeats the first pairing, the second pairing, and the second rational selection until there is no newly increased pairing-successful high intensity region pair.

(Appendix 6)
The device of clause 5, further comprising:
adding a virtual high-brightness region corresponding to the single high-brightness region of pairing failure, and substituting each high-brightness region pair consisting of said single high-brightness region and its corresponding said virtual high-brightness region into a third set of high-brightness regions; including additional units for pairing,
The second determining unit, based on the first set of high-brightness region pairs, the second set of high-brightness region pairs after a second rational selection, and the third set of high-brightness region pairs, An apparatus for determining the position of a vehicle in said input color image.

(Appendix 7)
A device according to Appendix 1, wherein
The first set of parameters includes at least one of a vertical coordinate value of the high-brightness region, an aspect ratio of a circumscribing rectangle, an area difference, a color difference, and a degree of shape matching;
The apparatus, wherein the second set of parameters includes at least one of a vertical coordinate value of the high intensity region, an aspect ratio of a bounding rectangle, and an area difference.

(Appendix 8),
An electronic device comprising the vehicle detection device according to any one of Appendices 1 to 7.

(Appendix 9)
A vehicle detection method comprising:
determining a high intensity region in the input color image;
expanding the high intensity region based on the color of the pixels surrounding each high intensity region;
After the expansion, the high-brightness regions with a predetermined color are paired for the first time, wherein each high-brightness region is selected based on the first set of parameters, and any two high-brightness regions are selected. Obtaining a first set of high-brightness region pairs by pairing each high-brightness region after selection based on the size difference between
Perform a second pairing for the high-brightness region of the first pairing failure, wherein the high-brightness region of the first pairing failure is selected based on the second set of parameters, and any obtaining a second pair of high-brightness regions by pairing each high-brightness region after selection based on the degree of shape matching between the two high-brightness regions; and determining a position of a vehicle in the input color image based on the intensity region pairs and the second set of high intensity region pairs.

(Appendix 10)
The method of Appendix 9, wherein
Expanding the high-brightness region based on the color of pixels surrounding each high-brightness region includes:
A method comprising expanding the high intensity regions to include pixels of a predetermined color around each high intensity region.

(Appendix 11)
The method of Supplementary Note 9, further comprising:
Before performing the first pairing, make a first rational selection for each high-brightness region after extension, among which, for each high-brightness region after extension, the area of the high-brightness region and / or making the selection based on the convexity of the high intensity regions;
Performing the first pairing on the high-brightness region having the predetermined color after the extension is the first pairing for the high-brightness region having the predetermined color after the first rational selection after the extension. A method comprising performing pairing.

(Appendix 12)
11. The method of Supplementary Note 11, further comprising:
After the second pairing, perform a second rational selection for the second set of high-brightness region pairs, wherein each high-brightness region pair in the second set of high-brightness region pairs , the color difference of the centroid, the luminance difference of the centroid, and the luminance histogram of the bounding rectangle, and merging based on the degree of overlap of each pair of high luminance regions. obtaining a second set of high intensity region pairs after the second rational selection with
Determining a position of a vehicle in the input color image based on the first set of high-brightness region pairs and the second set of high-brightness region pairs includes:
determining the position of the vehicle in the input color image based on the first set of high intensity region pairs and the second set of high intensity region pairs after a second reasonable selection; Method.

(Appendix 13)
The method of Supplementary Note 12, further comprising:
Repeating the first pairing, the second pairing, and the second rational selection a predetermined number of times, or the first pairing, the second pairing, and the second rational selection repeating the selective selection until there are no newly increased pairings of successful high intensity regions.

(Appendix 14)
13. The method of Supplementary Note 13, further comprising:
adding a virtual high-brightness region corresponding to the single high-brightness region of pairing failure, and substituting each high-brightness region pair consisting of said single high-brightness region and its corresponding said virtual high-brightness region into a third set of high-brightness regions; including pairing
Determining a position of a vehicle in the input color image based on the first set of high-brightness region pairs and the second set of high-brightness region pairs includes:
based on the first set of high-brightness region pairs, the second set of high-brightness region pairs after a second rational selection, and the third set of high-brightness region pairs in the input color image determining a position of a vehicle in a vehicle.

(Appendix 15)
The method of Appendix 9, wherein
The first set of parameters includes at least one of a vertical coordinate value of the high-brightness region, an aspect ratio of a circumscribing rectangle, an area difference, a color difference, and a degree of shape matching;
The method, wherein the second set of parameters includes at least one of a vertical coordinate value of the high intensity region, an aspect ratio of a bounding rectangle, and an area difference.

Although the preferred embodiment of the present invention has been described above, the present invention is not limited to this embodiment, and all modifications to the present invention fall within the technical scope of the present invention as long as they do not depart from the gist of the present invention.

Claims (10)

A vehicle detection device,
a first determining unit for determining a high brightness region in an input color image;
an expansion unit for expanding the high brightness region based on the color of the pixels surrounding each high brightness region;
A first pairing unit for performing a first pairing for the extended high-brightness regions having a predetermined color, wherein each high-brightness region is selected based on a first set of parameters. , a first pairing unit for obtaining a first pair of high-brightness regions by pairing each high-brightness region after selection based on the size difference between any two high-brightness regions; ;
A second pairing unit for performing a second pairing to the first pairing-failed high-brightness region, the second pairing unit for the first pairing-failed high-brightness region based on a second set of parameters. , and pairing each selected high-brightness region based on the degree of shape matching between any two high-brightness regions to obtain a second set of high-brightness region pairs. a second pairing unit; and a second determination unit for determining the position of the vehicle in the input color image based on the first set of bright region pairs and the second set of bright region pairs. apparatus, including A device according to claim 1, wherein
The apparatus of claim 1, wherein the expanding unit expands the high-brightness regions to include pixels of a predetermined color around each high-brightness region. The apparatus of claim 1, further comprising:
A first selection unit for performing a first selection for each high-brightness region after extension before performing the first pairing, wherein for each high-brightness region after extension, the high-brightness a first selection unit that selects based on the area of the region and/or the convexity of the high intensity region;
The device, wherein the first pairing unit performs the first pairing for the high-brightness region having a predetermined color after the first selection after expansion. 4. The apparatus of claim 3, further comprising:
a second selection unit for performing a second selection on the second set of high-brightness region pairs after performing the second pairing, wherein each of the second set of high-brightness region pairs making a selection based on at least one of the size of the high intensity region pair, the color difference of the centroid, the luminance difference of the centroid, and the luminance histogram of the bounding rectangle, and based on the degree of overlap of each high luminance region pair a second selection unit that obtains a second set of high-brightness region pairs after the second selection by merging;
The second determining unit determines the position of the vehicle in the input color image based on the first set of high brightness region pairs and the second set of high brightness region pairs after the second selection. ,Device. A device according to claim 4, wherein
The first pairing unit, the second pairing unit, and the second selection unit repeat the first pairing, the second pairing, and the second selection a predetermined number of times, or The apparatus repeats the first pairing, the second pairing, and the second selection until there is no newly increased pairing-successful high intensity region pair. 6. The apparatus of claim 5, further comprising:
adding a virtual high-brightness region corresponding to the single high-brightness region of pairing failure, and substituting each high-brightness region pair consisting of said single high-brightness region and its corresponding said virtual high-brightness region into a third set of high-brightness regions; including additional units for pairing,
The second determination unit is input based on the first set of high-brightness region pairs, the second set of high-brightness region pairs after the second selection, and the third set of high-brightness region pairs An apparatus for determining the position of a vehicle in said color image. A device according to claim 1, wherein
The first set of parameters includes at least one of a vertical coordinate value of the high-brightness region, an aspect ratio of a circumscribing rectangle, an area difference, a color difference, and a degree of shape matching;
The apparatus, wherein the second set of parameters includes at least one of a vertical coordinate value of the high intensity region, an aspect ratio of a bounding rectangle, and an area difference. An electronic device comprising the vehicle detection device according to any one of claims 1-7. A vehicle detection method executed by a vehicle detection device ,
determining high intensity regions in the input color image;
expanding the high intensity region based on the color of the pixels surrounding each high intensity region;
performing a first pairing of the enhanced regions having a predetermined color after the expansion, wherein for each enhanced region a selection is made based on a first set of parameters to select any two enhanced regions; obtaining a first set of high-brightness region pairs by pairing each high-brightness region after selection based on the size difference between the brightness regions;
performing a second pairing on the first pairing-failed high-brightness region, selecting the first pairing-failed high-brightness region based on a second set of parameters; obtaining a second set of high-brightness region pairs by pairing each high-brightness region after selection based on the degree of shape matching between any two high-brightness regions; and determining a position of a vehicle in said input color image based on a set of high intensity region pairs and said second set of high intensity region pairs. A method according to claim 9, wherein
Based on the color of pixels surrounding each high-brightness region, expanding the high-brightness region comprises:
A method comprising expanding the high intensity regions to include pixels of a predetermined color around each high intensity region.

Images