JP4347079B2 - Vehicle detection device and vehicle detection method - Google Patents

Description

  The present invention relates to a vehicle detection device and a vehicle detection method.

Conventionally, there has been a vehicle detection device that detects a vehicle traveling on a road using an imaging signal of a moving image or a still image captured by an imaging device such as a camera. As an example of such a vehicle detection device, for example, Patent Document 1 discloses a frame difference circuit that obtains a difference between image data of a current frame of an input video signal and image data of the previous frame, and the frame difference. And a profile circuit that performs profile processing that counts the number of pixels in the vertical and horizontal directions, and detects the vehicle width, vehicle leading edge, etc. based on the profile-processed data to detect the vehicle position and speed. A vehicle detection device for measuring the number of vehicles is described.
JP-A-5-250597

  However, in the above vehicle detection device, since the vehicle is detected by performing the frame difference process, it reacts to a moving part, for example, a part where the luminance has changed. In some cases, a shadow or a shadow of a vehicle traveling in the adjacent lane is erroneously detected as a traveling vehicle.

  The present invention has been made in view of the above-described conventional circumstances, and an object thereof is to provide a vehicle detection device and a vehicle detection method capable of detecting a vehicle with high accuracy and high speed.

The vehicle detection device of the present invention is
Scanning the image data, detecting a portion where the change in luminance value in the scanning direction of the image data is greater than or equal to a predetermined threshold as a peak point, and converting the image data into peak data including the position of the peak point A peak detection means for conversion;
First storage means for storing the peak data;
Subtracting the peak data output from the peak data creating means and the peak data stored in the first storage means at a time point before the output peak data, and subtracting the data A difference data creating means for creating difference data based on;
Second storage means for storing the difference data;
The difference data output from the difference data creating means and the difference data stored in the second storage means before the difference data are matched, and vehicle detection is performed based on the matching result. Vehicle determination means;
Is provided.

  With this configuration, it is possible to provide a vehicle detection device capable of detecting a vehicle with high accuracy and high speed.

  Further, the difference data creating means has projection processing means for projecting the data subjected to the subtraction process in the horizontal direction, and outputs the projected data as the difference data.

  With this configuration, since projection processing is performed to convert the data into one-dimensional data, the amount of data is compressed, and matching processing can be performed at high speed.

  The peak data is binary data indicating the presence or absence of the peak point.

  With this configuration, the input image data is converted into peak data, so that the amount of data can be compressed and the subsequent processing can be performed at high speed.

  Further, based on at least one of the number of the peak points in the area where the vehicle of the image data is imaged and the number of the peak points in the area where the road surface is imaged, the peak data creating means Peak determination threshold value control means for controlling the threshold value for determining peak data is further provided.

  With this configuration, the peak determination threshold value is controlled, so that highly accurate vehicle detection can be performed following the influence of changes in weather and illuminance.

Scanning the image data, creating peak data including a peak point position indicating a maximum or minimum value in a portion where the change in luminance value in the scanning direction of the image data is equal to or greater than a predetermined threshold;
Storing the peak data in a first storage means;
Creating difference data by performing a subtraction process between the created peak data and peak data at a time point earlier than the peak data stored in the first storage unit;
Storing the difference data in a second storage means;
Matching the difference data output from the difference data creation means with the difference data stored in the second storage means, the difference data at a time point before the difference data output from the difference data creation means;
Performing vehicle detection based on the matched results;
Is provided.

  With this configuration, it is possible to provide a vehicle detection device capable of detecting a vehicle with high accuracy and high speed.

  The step of creating the difference data includes a step of projecting the data subjected to the subtraction process in a horizontal direction and outputting the projected data as the difference data.

  With this configuration, since projection processing is performed to convert the data into one-dimensional data, the amount of data is compressed, and matching processing can be performed at high speed.

  The peak data is binary data indicating the presence or absence of the peak point.

  With this configuration, the input image data is converted into peak data, so that the amount of data can be compressed and the subsequent processing can be performed at high speed.

  Based on at least one of the number of the peak points in the area where the vehicle of the image data is imaged and the number of the peak points in the area where the road surface is imaged, the peak data of the peak data creating means The method further includes the step of controlling the threshold value for making the determination.

  With this configuration, the peak determination threshold value is controlled, so that highly accurate vehicle detection can be performed following the influence of changes in weather and illuminance.

  ADVANTAGE OF THE INVENTION According to this invention, the vehicle detection apparatus and vehicle detection method which can detect a vehicle with high precision and high speed can be provided.

  FIG. 1 is a schematic configuration diagram showing a vehicle detection device for explaining an embodiment of the present invention. As shown in FIG. 1, the vehicle detection device 1 according to the embodiment of the present invention includes a peak detection unit 10, a peak data storage unit 20, a frame difference processing unit 30, a projection processing unit 40, and a projection data storage unit 50. A vehicle determination unit 60, an output unit 70, and a peak determination threshold value control unit 80.

  The peak detector 10 scans the image data of the video signal for one frame input from the imaging device 2 in the horizontal direction. Then, a portion where the increase / decrease change of the scanned luminance value in the horizontal direction is equal to or greater than the peak determination threshold is detected as a peak point, and data including the position of the peak point (hereinafter, peak data) is output. The peak data output by the peak detection unit 10 is output to the frame difference unit 30 and also to the peak data storage unit 20.

  The peak data storage unit 20 stores the peak data output from the peak detection unit 10. The peak data storage unit 20 only needs to store peak data for at least one frame.

  The frame difference unit 30 reads the peak data one frame before the peak data output from the peak detection unit 10 from the peak data storage unit 20, the peak data output from the peak detection unit 10, and one frame before Is subtracted from the peak data and the difference data is output.

  The projection processing unit 40 generates and outputs projection data obtained by projecting the difference data output from the frame difference unit 30 in the horizontal direction. This projection data is, for example, a projection of the number of peaks in one horizontal line. The projection data output by the projection processing unit 40 is output to the vehicle determination unit 60 and also to the projection data storage unit 50.

  The projection data storage unit 50 stores the projection data output from the projection processing unit 40. The projection data storage unit 50 only needs to store at least one projection data (one time point) output from the projection processing unit 40.

  The vehicle determination unit 60 includes a matching unit 61 and a matching result determination unit 62. The matching unit 61 reads the projection data output from the projection data storage unit 50 one time before the projection data output from the projection processing unit 40 (before the temporary point), and the projection data output from the projection processing unit 40 And matching processing with the previously read projection data. An example of this matching method is DP (Dynamic Programming) matching, for example. The matching result determination unit 62 compares the result matched by the matching unit 61 with a predetermined matching threshold value to determine whether or not the vehicle has entered the imaging area.

  The output unit 70 outputs the determination result determined by the vehicle determination unit 60. The peak determination threshold control unit 80 controls the threshold for peak detection of the peak detection unit 10.

  FIG. 2 is a flowchart showing a vehicle detection method for explaining an embodiment of the present invention. First, as shown in FIG. 2, when image data is input from the imaging device 2 (step S201), the peak detector 10 scans the input image data in the horizontal direction to detect a peak point. Then, peak data is generated (step S202). The generated peak data is stored in the peak data storage unit 20 (step S203).

  Here, FIG. 3A and FIG. 3B are conceptual diagrams illustrating a captured image obtained by capturing the vehicle and the luminance value of the captured image on the horizontal line III. As shown in FIG. 3A, the vehicle 310 is captured in the image 300. Then, as shown in FIG. 3B, the luminance value in the horizontal line III is searched from the left side, and first, it is determined whether or not the luminance has changed by a predetermined peak determination threshold value TH. And when it changes more than the peak determination threshold value TH, the luminance value is further searched to the right side, and the point where the luminance change is maximum is set as a temporary peak point. In the example of FIG. 3 (b), a point A that is searched from the left side and increases by the threshold value TH and has the maximum luminance value is set as a temporary peak point.

  Then, the luminance value is further searched to the right side, and when the luminance value changes by more than the peak determination threshold TH from the temporary peak point, the temporary peak point is determined as the formal peak point. In the example of FIG. 3B, since the brightness TH has decreased by more than the luminance TH from the point A, the point A is determined as an official peak point.

  Next, a point at which the luminance change from the determined peak point is maximized is set as the next temporary peak point. In the example of FIG. 3B, the point B at which the luminance becomes minimum from the point where the luminance TH decreases from the point A is set as a temporary peak point.

  Then, the luminance value is further searched to the right side, and when the luminance value changes by more than the peak determination threshold value TH from the temporary peak point, the temporary peak point is determined as the formal peak point. In the example of FIG. 3B, since the luminance value increases from the point B to the right by the peak determination threshold value TH or more, the point B is determined as the formal peak point. Such peak detection is sequentially performed toward the right side, and when the horizontal line III is completed, peak detection is performed for other horizontal lines.

  Next, a case where a shadow of a vehicle traveling in the adjacent lane is imaged will be described. FIG. 4A and FIG. 4B are conceptual diagrams showing a captured image obtained by capturing the shadow of the vehicle and the luminance value of the captured image on the horizontal line IV. As shown in FIG. 4A, a shadow 410 of the vehicle is captured in the image 400. As shown in FIG. 4B, the luminance value in the horizontal line IV is searched from the left side, and first, a point D where the luminance has changed by more than the peak determination threshold TH is set as a temporary peak point. However, the point D is not detected as a peak point because the peak detection on the horizontal line IV ends from the point D without changing more than the peak determination threshold TH.

  Thus, even if the luminance value change caused by the shadow of the vehicle is greater than or equal to the peak detection threshold TH, it is determined that there is no peak point on the horizontal line IV of the image data. Therefore, in the case of processing with peak data, erroneous detection due to the shadow of the vehicle can be prevented by converting the image data that is the basis of vehicle determination into peak data.

  The input image data is, for example, multivalued image data, and the generated peak data is, for example, binary data indicating a position and the presence or absence of a peak at that position. In this way, by converting the input multi-valued image data into binary peak data, subsequent processing can be performed at high speed. The position information included in the peak data may be indicated for each pixel in the horizontal direction, or the horizontal direction may be divided into a predetermined number of blocks, and the presence or absence of a peak in the block may be determined.

  Returning to FIG. 2, the description of the vehicle detection method of the present embodiment will be continued. When the peak is detected by the peak detector 10, the frame difference unit 30 subtracts the peak data output from the peak detector 10 and the peak data of the previous frame stored in the peak data storage unit 20. To generate difference data (step S204).

  Then, the projection processing unit 40 performs projection processing for projecting the difference data in the horizontal direction to generate projection data (step S205). The generated projection data is stored in the projection data storage unit 50 (step S206).

  5 and 6 are conceptual diagrams showing image input, frame difference, and projection processing. FIG. 5 shows a case where a vehicle is imaged, and FIG. 6 shows a case where a shadow such as a roadside tree is imaged. Yes.

  In FIG. 5, images 501, 502, and 503 are images for each frame, image 502 is an image after one frame of image 501, and image 503 is an image after one frame of image 502. The images 501 to 503 are detected as peaks and converted into peak data.

  Then, the image 501 and the image 502 converted into the peak data, the image 502 and the image 503 are subtracted, respectively, and become differential data 510 and 511. Then, projection processing is performed for each horizontal line in the difference data 510 and 511, and projection data 520 and 521 are obtained. Here, this projection process is a process of integrating the number of peak points for each horizontal line, for example.

  Similarly in FIG. 6, the images 601, 602, and 603 are images for each frame, the image 602 is an image that is one frame after the image 601, and the image 603 is an image that is one frame after the image 602. The images 601 to 603 are detected as peaks and converted into peak data. Then, the image 601 and the image 602 converted into the peak data, the image 602 and the image 603 are subtracted, respectively, to become difference data 610 and 611. Then, projection processing is performed for each horizontal line in the difference data 610 and 611 to obtain projection data 620 and 621.

  Thus, since the data amount is reduced by converting the difference data into one-dimensional projection data, the subsequent processing can be performed at high speed.

  Returning to FIG. 2 again, the description of the vehicle detection method of this embodiment will be continued. When the projection data is output by the projection processing unit 40, the matching unit 61 of the vehicle determination unit 60 calculates the projection data input from the projection processing unit 40 and the previous projection data stored in the projection data storage unit 50. Matching is performed to obtain the degree of coincidence (step S207).

  Then, the matching result determination unit 62 compares the matching degree obtained by the matching unit 61 with the matching threshold value (MTG) (step S208). If the degree of coincidence is greater than a predetermined threshold, it is determined that there is a vehicle (step S209), and if it is equal to or lower than the predetermined threshold, it is determined that there is no vehicle (step S210).

  That is, in FIG. 5, when the vehicle is imaged, the projection data 520 and 521, which is data obtained by taking a frame difference, does not change the vehicle speed significantly for each frame. The shapes of these are almost the same.

  On the other hand, since the shadow of the roadside tree moves randomly, the projection data 620 and 621 which are data obtained by taking a frame difference also change randomly, and the degree of coincidence of the shapes of the projection data 620 and 621 is low. .

  Thus, since matching is performed between the projection data and the vehicle is detected based on the degree of coincidence, the vehicle can be detected with high accuracy without erroneously detecting the shadow of the roadside tree.

  Note that it is also possible to determine the vehicle head position of the imaged vehicle based on this projection data, and to determine the traveling direction, traveling speed, etc. of the vehicle.

  Next, control of the peak determination threshold value TH for peak detection will be described. In the captured image data, the vehicle detection device of the present embodiment detects the peak point and converts it to peak data, and performs vehicle detection by taking the difference of the peak data for each frame, so the road surface in the image data is It is preferable that the peak in the imaged portion is not detected as much as possible.

  However, peak detection is affected even on the same road surface due to the environment of the place to be imaged, changes in weather, changes in illuminance such as street lights, and the like. Therefore, the vehicle detection apparatus of the present embodiment controls the peak determination threshold value TH (FIG. 3) using the peak determination threshold value control unit 80, thereby suppressing the influence of peak detection.

  Here, when the peak determination threshold value TH is small, a fine change in luminance value of a portion where an object other than the vehicle is imaged, such as a road surface, is also detected as a peak. Therefore, the peak determination threshold needs to be increased to some extent.

  On the other hand, when the peak determination threshold value TH is large, instead of detecting the change in the luminance value of the part where the vehicle is imaged as a peak, the change in the luminance value of the part where the vehicle is imaged is also detected as a peak. It will disappear.

  Therefore, since the change in the luminance value of the portion where the vehicle is imaged is larger than the change in the luminance value of the portion where the vehicle other than the vehicle such as the road surface is imaged, the peak determination threshold control unit 80 captures the vehicle. The peak determination threshold value is controlled so that no peak exists in the portion where the vehicle is imaged, and there exists a peak in the portion where the vehicle is imaged.

  Specifically, the peak determination threshold value control unit 80 reads the peak data from the peak data storage unit 20, and captures a portion where the vehicle is imaged (hereinafter referred to as a vehicle area) and a portion where images other than the vehicle such as the road surface are captured ( Hereinafter, a road surface area) is detected. The vehicle area and the road surface area are detected by, for example, projecting peak data in the horizontal direction, detecting the head position of the portion where the peak is continuously detected in the vertical direction as the vehicle head, and the vehicle area based on the vehicle head position. And detect the road surface area.

  Then, the peak determination threshold value control unit 80 measures the number of peak points in the road surface area and the number of peak points in the vehicle area. When the number of peak points in the road surface area is greater than a predetermined value, the peak determination threshold value is increased to reduce the peak detection sensitivity in the road surface area. On the other hand, when the number of peak points in the vehicle area is less than a predetermined value, the peak determination sensitivity is lowered to increase the peak detection sensitivity in the vehicle area.

  Note that the peak determination threshold value control unit 80 may calculate the peak determination threshold value based on a predetermined function from the number of measured peak points, or create a peak determination threshold value control table. You may memorize | store and obtain | require with reference to the table from the number of the measured peak points.

  Moreover, the peak determination threshold value control unit 80 may control the peak determination threshold value TH for each vehicle based on the vehicle determination result from the matching result determination unit 62. In that case, the control may be performed for each vehicle, and the peak determination threshold value may be changed at the time when the vehicle is not imaged after passing the vehicle.

  In this way, by controlling the peak determination threshold value TH, it is possible to perform highly accurate vehicle detection following the influence of changes in weather and illuminance.

  According to the vehicle detection device and the vehicle detection method of the embodiment of the present invention as described above, since the frame difference is performed by detecting the peak of the input video data, it depends on the shadow of the vehicle traveling in the adjacent lane. False detection can be avoided. Further, since the amount of data can be compressed by converting the input video data into peak data, high-speed processing becomes possible.

  Moreover, since the data which performed the frame difference are matched, it can be determined whether the motion of the detected thing is a vehicle, and the misdetection by the shadow of a roadside tree, etc. can be avoided. Furthermore, since the data subjected to the frame difference is converted into one-dimensional data by performing a projection process, the matching process can be performed at high speed.

  In the present embodiment, an example in which the imaging device and the vehicle detection device are separately provided has been described, but these may be integrally configured.

  Further, although the case where the vehicle detection process is performed for each frame has been described as an example, it is not necessarily performed for each frame. For example, depending on the speed limit of the road on which the vehicle is detected, for example, every predetermined time or every predetermined frame As described above, the frequency of processing may be reduced.

  The present invention has an effect of being able to detect a vehicle with high accuracy and high speed, and is useful for a vehicle detection device, a vehicle detection method, and the like.

The schematic block diagram which shows the vehicle detection apparatus for describing embodiment of this invention The flowchart which shows the vehicle detection method for describing embodiment of this invention Conceptual diagram showing a captured image obtained by capturing the vehicle and a luminance value in horizontal line III of the captured image The conceptual diagram which shows the luminance value in the horizontal line IV of the captured image by which the shadow of the vehicle was imaged, and the captured image Conceptual diagram showing image input, frame difference, and projection processing Conceptual diagram showing image input, frame difference, and projection processing

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Vehicle detection apparatus 2 Imaging apparatus 10 Peak detection part 20 Peak data storage part 30 Frame difference part 40 Projection process part 50 Projection data storage part 60 Vehicle determination part 61 Matching part 62 Matching result determination part 70 Output part

Claims (8)

Scanning the image data, the change in increase and decrease of luminance values in the scanning direction of the image data to detect a peak point showing the maximum or minimum value in a portion not less than a predetermined threshold value, the peak point of the image data Peak detection means for converting into peak data including the position of
First storage means for storing the peak data;
Based on the data obtained by subtracting the peak data output from the peak detection means and the peak data stored in the first storage means at a time point before the output peak data. Differential data creation means for creating differential data,
Second storage means for storing the difference data;
Based on the degree of coincidence between the difference data by matching the difference data output from the difference data creating means with the difference data at a time point before the difference data stored in the second storage means Vehicle determination means for performing vehicle detection;
A vehicle detection device comprising:
  The vehicle detection device according to claim 1, wherein the difference data creating unit includes a projection processing unit that projects the data subjected to the subtraction process in a horizontal direction, and outputs the projected data as the difference data.   The vehicle detection device according to claim 1, wherein the peak data is binary data indicating the presence or absence of the peak point.   Peak data determination of the peak detection means based on at least one of the number of the peak points of the area where the vehicle of the image data is imaged and the number of the peak points of the area where the road surface is imaged The vehicle detection device according to any one of claims 1 to 3, further comprising a peak determination threshold value control means for controlling the threshold value for performing the operation. Scanning image data, and generating peak data including a peak point position indicating a maximum value or a minimum value in a portion where a change in luminance value in the scanning direction of the image data is equal to or greater than a predetermined threshold value ; ,
Storing the peak data in a first storage means;
Subtracting the created peak data from the peak data stored in the first storage means before the peak data to create difference data;
Storing the difference data in a second storage means;
Matching the created difference data with the difference data stored in the second storage means and at a time point before the difference data output from the difference data creating means;
Performing vehicle detection based on the degree of coincidence between the difference data;
A vehicle detection method comprising:   The vehicle detection method according to claim 5, wherein the step of creating the difference data includes a step of projecting the data subjected to the subtraction process in a horizontal direction and outputting the projected data as the difference data.   The vehicle detection method according to claim 5 or 6, wherein the peak data is binary data indicating the presence or absence of the peak point.   Based on at least one of the number of the peak points in the area where the vehicle of the image data is imaged and the number of the peak points in the area where the road surface is imaged, the peak data of the peak data creating means The vehicle detection method according to any one of claims 5 to 7, further comprising a step of controlling the threshold value for making the determination.

Images