JP2013101534A - Vehicle driving support system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform efficient Hough voting and to increase a processing speed of traveling lane detection.SOLUTION: From an image picked up by an imaging part 10, a tail lamp of a preceding vehicle traveling right in front of a present vehicle is detected through a spectral filter by a tail lamp detection part 32. Then, a position of the preceding vehicle is detected on the basis of the image of the detected tail lamp. By a white line detection part 33, a candidate edge of a traveling lane of a road surface is extracted on the basis of the position of the preceding vehicle, Hough transformation processing is executed to the extracted candidate edge, and the traveling lane is determined.

Description

  In the present invention, when driving a vehicle such as an automobile, it is monitored whether or not the vehicle is likely to depart from a traveling lane defined by a traveling lane (including white and yellow lines), and an alarm is sounded if the vehicle is likely to deviate. The present invention relates to a vehicle driving support system that performs driving to support driving of the vehicle.

  As a system used for driving assistance (driver assistance) of an automobile, a system based on object recognition on an image in front of the host vehicle taken by an in-vehicle camera is known. Specifically, there is an LDW (Lane Departure Warning) system that monitors whether or not the own vehicle is likely to deviate from the traveling lane photographed by the in-vehicle camera and sounds an alarm if the vehicle is likely to deviate. In this LDW system, a road image in front of the host vehicle is captured by an imaging unit of an in-vehicle camera mounted on the vehicle, and the traveling lane is detected by utilizing the high luminance of the traveling lane. Specifically, for example, an edge (edge point sequence) is extracted by image differentiation processing and binarization processing, and it is determined whether the extracted edge is a traveling lane edge. It is already known that the position of the traveling lane which is the traveling lane is detected by performing a Hough transform as a straight line detection method on the extracted edge.

  As this traveling lane detection method, the method described in Patent Document 1 is known. In the traveling lane detection method of Patent Document 1, preprocessing is performed on an image input by an in-vehicle camera, and the calculated candidate points (pixels) of the traveling lane are stored in a storage area of a memory to which image coordinates are assigned. Vote. And the pixel with the largest number of Hough votes is recognized as a travel lane because it is the most frequent straight line on the input image.

  However, in the traveling lane detection method of Patent Document 1, the Hough voting is performed on all image regions of the input image, so the number of Hough voting increases and the Hough conversion process becomes heavy. For this reason, the processing speed of travel lane detection falls. Such a decrease in processing speed is a major obstacle for the LDW system that requires real-time processing.

  The present invention has been made in view of the above problems, and an object of the present invention is to provide a vehicle driving support system capable of performing efficient driving lane detection processing without waste and increasing the processing speed of driving lane detection. That is.

  In order to achieve the above object, the invention of claim 1 comprises imaging means for imaging the front of the host vehicle, and traveling lane detecting means for detecting the driving lane on the road surface in the image captured by the imaging means, In a vehicle driving support system that monitors whether the host vehicle is likely to deviate from the detected driving lane and supports driving of the vehicle, the position of the preceding vehicle that is traveling immediately in front of the host vehicle from the image captured by the imaging unit is determined. A forward vehicle position detecting means for detecting is provided, and a traveling lane is detected by the traveling lane detecting means based on the position of the forward vehicle detected by the forward vehicle position detecting means.

  In the present invention, since it is highly possible that the forward vehicle traveling immediately before the host vehicle is traveling in the traveling lane of the host vehicle, the traveling lanes are respectively displayed in the lower left and lower right regions of the preceding vehicle on the image. Exists. By processing the lane detection target image area from the position of the vehicle ahead in the image captured by the imaging means to the lower left and lower right image areas, processing of the traveling lane detection process performed on all images captured by the imaging means It can be reduced compared to the number. As a result, the traveling lane detection process becomes faster, and the traveling lane detection processing speed increases.

  According to the present invention, it is possible to obtain a unique effect of performing efficient travel lane detection processing without waste and increasing the processing speed of travel lane detection.

It is a block diagram which shows the outline | summary of the whole structure of the vehicle driving assistance system which concerns on this embodiment. It is a block diagram which shows the structure of the signal processing part in this vehicle driving assistance system. It is a figure explaining the positional information on the tail lamp of a front vehicle. It is a flowchart which shows the left lane detection operation | movement of the Hough vote object pixel determination process in this vehicle driving assistance system. It is a flowchart which shows the right lane detection operation | movement of the Hough vote object pixel determination process in this vehicle driving assistance system. It is the schematic which shows the scanning direction of a Hough vote object pixel.

Hereinafter, an embodiment in which the present invention is applied to a vehicle driving support system that supports driving by driving an automobile along a driving lane (white line) will be described with a focus on a configuration related to detection of a white line on a road.
FIG. 1 is a block diagram showing an outline of the overall configuration of the vehicle driving support system according to the present embodiment. FIG. 2 is a block diagram illustrating a configuration of the signal processing unit.
The vehicle driving support system includes an imaging unit 10 and a signal processing unit 20. The imaging unit 10 extracts light having a specific wavelength out of the light collected by the lens 11 by emphasizing it with the spectral filter 12, and photoelectrically converts the extracted light with the sensor 13 to output image data. Here, the spectral filter 12 is used in order to accurately detect the tail lamp (red light) of the front vehicle traveling in front of the host vehicle and to acquire image data of the tail lamp. The signal processing unit 20 includes an FPGA 21 including a CPU as a signal processing circuit, and a memory 22 that is a frame buffer for storing image data. The signal processing unit 20 includes a spectral image generation unit 31, a tail lamp detection unit 32, and a white line detection unit 33. The white line detection unit 33 includes an edge detection unit 33-1, a binarization unit 33-2, a Hough vote target pixel determination unit 33-3, a Hough conversion unit 33-4, and a Hough inverse conversion unit 33-5. Has been. In the present embodiment, the Hough transform process, which is one of the white line straight line detection methods, is used, but the present invention is not limited to this.

  The tail lamp detection unit 32 detects a tail lamp from the spectral image and the monochrome image generated by the spectral image generation unit 31 from the image data received from the imaging unit, and the position coordinates thereof are the Hough voting target pixel determination unit of the white line detection unit 33. Output to 33-3. In the white line detection unit 33, the input monochrome image is subjected to edge detection by the edge detection unit 33-1, and then binarized by the binarization unit 33-2. After that, the Hough voting target pixel determination unit 33-3 performs a Hough voting target pixel determination process that removes pixels other than the Hough voting target pixels from the binarized image. The Hough transform unit 33-4 performs Hough transform on the image from which unnecessary pixels are removed, and the Hough inverse transform unit 33-5 calculates and outputs the coordinates of the straight line having the highest appearance frequency in the Hough inverse transform. . In this embodiment, pixels unnecessary for the Hough voting are removed using the position coordinates of the tail lamp before the Hough conversion, and the efficient Hough voting is performed. The position information of the tail lamp is a value as shown in FIG. For example, the image captured by the imaging unit 10 stored in the memory 22 is subjected to image processing, whereby the width w and height h of the area including the tail lamp of the vehicle ahead are measured by the number of pixels. Then, the position coordinates (cx, cy) of the center of the tail lamp are calculated based on the measured width and height.

  Next, the Hough voting target pixel determination process in the present embodiment will be described with reference to FIGS. 4 and 5 showing the process flow. FIG. 6 is a schematic diagram showing the scanning direction of the Hough voting target pixel. In the rear image of the front vehicle in FIG. 6, the upper left is the origin (0,0), and the x coordinate is positive in the downward direction on the image and the y coordinate is positive in the right direction on the image. Shall be shown. Here, the Hough voting target pixel determination process is a process of leaving a pixel to be subjected to Hough voting (white pixel) and removing a pixel not to be performed (black pixelation) from the binarized image. When the tail lamp cannot be detected at the time of determining the Hough voting target pixel, pixels that are white line candidates are scanned from the preset scan start position for the binarized image to determine the Hough voting pixel.

  First, according to the Hough voting target pixel determination process for detecting the left lane shown in FIG. 4, a predetermined image area (width threshold (WIDTH) and height threshold (HEIGHT)) is defined for an image captured by the imaging unit. (Image area surrounded by). After edge detection is performed on the clipped image, a binarized binarized image is input (step S101). Then, the position information of the tail lamp (red light) of the preceding vehicle included in the input binarized image is acquired (step S102). The position information of the tail lamp is the center coordinates (cx, cy), width (w), and height (h) of the area including the tail lamp, and is detected by measuring the dispersed pixels in the binarized image. it can. Next, in the cut out image area, pixels in the image area surrounded by the width of WIDTH and the height of (cy + w) are replaced with black pixels (step S103). The scanning start point (Lx, Ly) for detecting the left lane is a predetermined length directly below the center coordinate position of the area including the tape lamp (here, it is w, but it is not necessary to limit to w) The coordinate (cx, cy + w) moved by the minute is set (step S104). Then, the white pixel discovery flag is initialized (step S105). A search for white pixels is started, and scanning is performed when the x coordinate of the scanning point is from 0 to (cx−w / 2) (steps S106 and S107). If there is a white pixel, if the white pixel detection flag is 0, a Hough vote is performed by setting the flag to 1 (step S108: YES, step S109: YES, step S110). Next, the x coordinate of the scanning point is incremented by 1, and the process returns to step S106 to repeat the white pixel scanning (step S111). If the x-coordinate of the scanning point exceeds (cx−w / 2) in step S107, the white pixel in the exceeded area is replaced with a black pixel to be a white line candidate point. The x coordinate of the scanning point is incremented by 1, and the process returns to step S106 to repeat white pixel scanning (step S107: NO, step S112, step S111). Even in the case where no white pixel is found as a result of scanning in step S108, the x coordinate of the scanning point is increased by 1, and the process returns to step S106 to repeat the white pixel scanning (step S108: NO, step S111). When the x coordinate of the scanning point is 0 or less When the y coordinate of the scanning point is smaller than the height threshold (HEIGHT), the scanning point is set to the x coordinate of the center position of the area including the tape lamp. After the y coordinate is increased by 1 and the white pixel detection flag is set to 0, the process returns to step S106 to repeat the white pixel scanning (step S106: NO, step S113: YES, step S114). If the y coordinate of the scanning point is equal to or higher than the height threshold (HEIGHT) in step S113, the Hough voting target pixel determination process according to the present embodiment is terminated (step S113: NO). Here, when the tail lamp cannot be detected, pixels that are white line candidates are scanned from a preset scan start position for the binarized image, and a Hough voting pixel is determined.

  Next, according to the Hough voting target pixel determination process for detecting the right lane shown in FIG. 5, a predetermined image area (width threshold (WIDTH) and height threshold (HEIGHT)) for the image captured by the imaging unit. (Image area surrounded by). After edge detection is performed on the clipped image, a binarized binary image is input (step S201). Then, the position information of the tail lamp (red light) of the preceding vehicle included in the input binarized image is acquired (step S202). The position information of the tail lamp is the center coordinates (cx, cy), width (w), and height (h) of the area including the tail lamp, and is detected by measuring the dispersed pixels in the binarized image. it can. Next, in the clipped image area, pixels in the image area surrounded by the width of WIDTH and the height of (cy−w) are replaced with black pixels (step S203). Then, the scanning start point (Lx, Ly) for detecting the left lane is set to the coordinates (cx, cy + w) moved by the length w from the center coordinate position of the area including the tape lamp (step S204). . Then, the white pixel discovery flag is initialized (step S205). A search for white pixels is started, and scanning is performed between the x coordinate of the scanning point from the threshold value (WIDTH) to (cx + w / 2) (steps S206 and S207). If there is a white pixel, if the white pixel detection flag is 0, a Hough vote is performed by setting the flag to 1 (step S208: YES, step S209: YES, step S210). Next, the x coordinate of the scanning point is incremented by 1, and the process returns to step S206 to repeat the white pixel scanning (step S211). If the x-coordinate of the scanning point exceeds (cx + w / 2) in step S207, the white pixel in the exceeded area is replaced with a black pixel to be a white line candidate point. The x coordinate of the scanning point is increased by 1, and the process returns to step S206 to repeat the white pixel scanning (step S207: NO, step S212, step S211). Even in the case where no white pixel is found as a result of scanning in step S208, the x coordinate of the scanning point is incremented by 1, and the process returns to step S206 to repeat the white pixel scanning (step S208: NO, step S211). When the x coordinate of the scanning point is equal to or smaller than the width threshold (WIDTH) When the y coordinate of the scanning point is smaller than the height threshold (HEIGHT), the x coordinate of the scanning point is set to x of the center position of the area including the tape lamp. In the coordinates, the y coordinate of the scanning point is incremented by 1 and the white pixel finding flag is set to 0, and then the process returns to step S106 to repeat the white pixel scanning (step S106: NO, step S113: YES, step S114). If the y coordinate of the scanning point is equal to or higher than the height threshold (HEIGHT) in step S113, the Hough voting target pixel determination process according to the present embodiment is terminated (step S113: NO). Here, when the tail lamp cannot be detected, pixels that are white line candidates are scanned from a preset scan start position for the binarized image, and a Hough voting pixel is determined.

  In the above-described embodiment, the position of the vehicle ahead is detected by detecting the tail lamp of the vehicle ahead in the travel lane of the host vehicle, and the Hough vote is efficiently performed by limiting the pixels to perform the Hough vote based on the position information. Hough conversion. The tail lamp of the preceding vehicle is detected to detect the position of the preceding vehicle. However, the present invention is not limited to this. For example, the same effect as in the above embodiment can be obtained by detecting the tread of the rear tire of the preceding vehicle. It is done. In this case, the tread can be detected with high accuracy by performing processing using a polarizing filter or the like for removing the reflection component from the road surface, and the same Hough voting as in the above embodiment is obtained from the center coordinates of the treads of both rear wheels. The target pixel determination process is performed. In addition, the same effect as that of the above embodiment can be obtained by detecting an automobile registration number mark (number plate) mainly installed at the center of the vehicle. In this case, the spectral registration process acquires white image data of a predetermined area, green image of a business vehicle, or yellow image data for a light vehicle, and detects the position of the vehicle registration number mark. Hough voting target pixel determination processing similar to that in the above embodiment is performed from the center coordinates of the position.

What has been described above is merely an example, and the present invention has a specific effect for each of the following modes.
(Aspect A)
Based on the position of the forward vehicle detected by the forward vehicle position detecting means, comprising forward vehicle position detecting means for detecting the position of the forward vehicle traveling immediately in front of the host vehicle from the image captured by the imaging means, The travel lane is detected by the travel lane detection means. According to this, as described in the above embodiment, there is a high possibility that a vehicle traveling immediately in front of the host vehicle is traveling in the traveling lane of the host vehicle. There is a driving lane in the lower right area. The position of the front vehicle is detected by detecting a tail lamp in the rear part of the front vehicle. Then, by limiting the lower left and lower right image areas from the detected position of the preceding vehicle to the image area to be detected as the traveling lane, the number of Hough votes in the Hough transform, which is one of the traveling lane detection processes, is reduced. Thereby, the processing speed of traveling lane detection can be raised.
(Aspect B)
In (Aspect A), the imaging means electrically converts an optical system including a lens, a spectral filter that separates imaging light that has passed through the optical system into spectral components, and light of a desired spectral component that has passed through the spectral filter. And a sensor that converts the signal into a signal. According to this, as described in the above embodiment, it is possible to easily acquire the position information of the preceding vehicle by detecting the image region including the red tail lamp of the preceding vehicle via the spectral filter 12, and the traveling The processing speed of lane detection can be increased.
(Aspect C)
In (Aspect A) or (Aspect B), the traveling lane detecting means extracts candidate edges of the road lane on the road surface by binarization processing of the image captured by the imaging means, and performs Hough transform on the extracted candidate edges. To detect the driving lane. According to this, as described in the above embodiment, the position of the forward vehicle can be detected more easily, so that the Hough conversion process is accelerated and the processing speed of the traveling lane detection can be increased.
(Aspect D)
In (Aspect A), the signal processing means includes spectral image generation means for generating a spectral image from the image data from the imaging means. According to this, as described in the above embodiment, the position of the vehicle ahead can be detected more easily by obtaining an image corresponding to the tail lamp from the image captured by the spectral image generation unit 31. Therefore, the Hough conversion process is accelerated, and the traveling lane detection process speed can be increased.
(Aspect E)
In (Aspect A), the forward vehicle position detection means includes tail lamp detection means for detecting the tail lamp of the forward vehicle. According to this, as described in the above embodiment, by detecting the red tail lamp of the vehicle ahead, the Hough conversion process is accelerated, and the traveling lane detection processing speed can be increased.
(Aspect F)
In (Embodiment E), the coordinates of the tail lamp of the preceding vehicle detected by the tail lamp detection means are obtained, and the candidate edges of the road lane on the road surface are extracted based on the coordinates of the tail lamp by the travel lane detection means. According to this, as described in the above embodiment, by limiting the image area to be subjected to the Hough conversion based on the coordinates of the tail lamp of the preceding vehicle, the Hough conversion process becomes faster, and the traveling lane detection processing speed is increased. Can be raised.
(Aspect G)
In (Aspect E) or (Aspect F), when the tail lamp of the vehicle ahead cannot be detected by the tail lamp detection means, the binarized image subjected to the binarization processing of the image picked up by the image pickup means is displayed on the road surface. A candidate edge of the road lane on the road surface is extracted from a preset scan start position for extracting a candidate edge of the lane. According to this, as described in the above embodiment, even if the position of the preceding vehicle cannot be detected, pixels that are travel lane candidates are scanned from the preset scan start position with respect to the binarized image, and Voting pixels can be determined.
(Aspect H)
In (Aspect A), the forward vehicle position detection means includes a tread detection means for detecting a tread of a rear wheel tire of the forward vehicle. According to this, as described in the modification of the above embodiment, the processing speed of the traveling lane detection can be increased by detecting the tread of the rear wheel tire of the front vehicle and detecting the position of the front vehicle. .
(Aspect I)
In (Aspect A), the forward vehicle position detection means includes automobile registration number mark detection means for detecting the automobile registration number mark of the preceding vehicle. According to this, as described in the modification of the above-described embodiment, the processing speed of the traveling lane detection can be increased by detecting the vehicle registration number mark of the preceding vehicle and detecting the position of the preceding vehicle.

DESCRIPTION OF SYMBOLS 10 Image | photographing part 11 Lens 12 Spectral filter 13 Sensor 20 Signal processing part 21 FPGA
22 Memory 31 Spectral Image Generation Unit 32 Tail Lamp Detection Unit 33 White Line Detection Unit 33-1 Edge Detection Unit 33-3 Binarization Unit 33-3 Hough Voting Target Pixel Determination Unit 33-4 Hough Conversion Unit 33-5 Hough Inverse Conversion Unit

JP 2007-179386 A

Claims (9)

An imaging unit that images the front of the host vehicle and a traveling lane detection unit that detects a traveling lane on the road surface in an image captured by the imaging unit, and monitors whether the host vehicle is likely to deviate from the detected traveling lane. In a vehicle driving support system that supports driving,
Forward vehicle position detection means for detecting the position of a forward vehicle traveling immediately in front of the host vehicle from an image captured by the imaging means;
A vehicle driving support system, wherein a traveling lane is detected by the traveling lane detecting means based on a position of a preceding vehicle detected by the preceding vehicle position detecting means. The vehicle driving support system according to claim 1,
The imaging means includes an optical system including a lens, a spectral filter that separates imaging light passing through the optical system for each spectral component, and a sensor that converts light of a desired spectral component that passes through the spectral filter into an electrical signal. A vehicle driving support system comprising: The vehicle driving support system according to claim 1 or 2,
The travel lane detecting means extracts a candidate edge of the road lane on the road surface by binarization processing of the image captured by the imaging means, and performs a Hough transform process on the extracted candidate edge to detect the travel lane. A vehicle driving support system. The vehicle driving support system according to claim 1,
The vehicle driving support system, wherein the traveling lane detecting unit includes a spectral image generating unit that generates a spectral image from image data from the imaging unit. The vehicle driving support system according to claim 1,
The vehicle front support system according to claim 1, wherein the front vehicle position detection means includes tail lamp detection means for detecting a tail lamp of a front vehicle. The vehicle driving support system according to claim 5, wherein
A vehicle driving support system characterized in that coordinates of a tail lamp of a preceding vehicle detected by the tail lamp detecting means are obtained, and candidate edges of a road lane on the road surface are extracted based on the coordinates of the tail lamp by the traveling lane detecting means. The vehicle driving support system according to claim 5 or 6,
When the tail lamp of the vehicle ahead cannot be detected by the tail lamp detection means, candidate edges of road lanes on the road surface are extracted from the binarized image subjected to the binarization processing of the image captured by the imaging means. A vehicle driving support system that extracts a candidate edge of a road lane on a road surface from a preset scan start position for the purpose. The vehicle driving support system according to claim 1,
The front vehicle position detection means includes a tread detection means for detecting a tread of a rear tire of a front vehicle. The vehicle driving support system according to claim 1,
The vehicle front support system according to claim 1, wherein the front vehicle position detection means includes a car registration number mark detection means for detecting a car registration number mark of a preceding vehicle.

Images