JP2008165595A - Obstacle detection method, obstacle detection device, obstacle detection system - Google Patents

Abstract

An obstacle detection method, an obstacle detection device, and an obstacle detection system capable of accurately detecting an obstacle such as a pedestrian regardless of summer or daytime and reducing the processing amount as compared with the conventional one.
An LSI detects a plurality of edge points in one captured image, and divides each area surrounding each detected edge point into blocks. The LSI 31 calculates the correlation value by searching the other captured image using the edge area including the edge point as a reference area among the divided blocks. The LSI 31 specifies the region having the largest correlation value as the corresponding region, and calculates the parallax between the edge region and the corresponding region. The LSI 31 specifies a candidate area having substantially the same distance from the imaging device 1 to the imaging target based on the calculated parallax. The LSI 31 performs the above processing for each edge region to identify a plurality of candidate regions, extracts an obstacle candidate region composed of the identified plurality of candidate regions, and based on the extracted feature amount of the obstacle candidate region To detect obstacles.
[Selection] Figure 3

Description

  The present invention relates to an obstacle detection method, an obstacle detection device, and an obstacle for detecting an obstacle such as a pedestrian and a parked vehicle by processing a captured image obtained by imaging the periphery (for example, the front) of the vehicle. The present invention relates to an obstacle detection system including an object detection device.

  A video camera sensitive to far infrared rays is mounted on a vehicle such as an automobile, the image captured by the video camera is binarized, the distance to the binarized object is calculated, and a pedestrian in front of the vehicle, A vehicle periphery monitoring device that extracts animals and the like has been proposed (see Patent Document 1).

Also, one of the images captured by the two cameras is divided into blocks of a certain size, the other image (reference image) is searched using each block as a reference area, and a correlation value is calculated. There has been proposed an apparatus for extracting a pedestrian by extracting a corresponding area, calculating a parallax between a reference area and the corresponding area, and grouping equidistant blocks (see Patent Document 2).
JP 2004-135034 A Japanese Patent Laid-Open No. 05-114099

  In the vehicle periphery monitoring device of Patent Document 1, on the assumption that the temperature of the pedestrian or animal is higher than that of the vehicle periphery, a pedestrian or the like is performed by setting a binarization threshold and performing binarization processing. Is extracted. However, in summer or in the daytime, the temperature of surrounding buildings is often higher than that of pedestrians or animals, and when binarization processing is performed, the pedestrian or animal has a luminance that is higher than the binarization threshold. Since it does not have, there exists a problem that a pedestrian or an animal cannot be extracted accurately.

  Further, in the apparatus of Patent Document 2, when searching for the other image using a block as a reference area, it is necessary to perform search processing for the block of the entire image, and the amount of processing such as calculation of correlation values, calculation of parallax, etc. There is a problem that becomes enormous. In addition, when one image is divided into blocks, the image is simply mechanically divided. Therefore, when a block having a uniform luminance and a small feature in the block is used as a reference area, Therefore, there is a possibility that a plurality of corresponding area candidates may be extracted, and there is a problem that it is impossible to accurately extract a pedestrian.

  The present invention has been made in view of such circumstances, and an edge point is detected from one captured image captured by two imaging devices, and a block (small region) including the edge point is used as a reference region. Pedestrians and animals regardless of summer or daytime by searching the other captured image (reference image), identifying the corresponding area, and grouping equidistant blocks based on the parallax between them to detect obstacles An obstacle detection method, an obstacle detection device, and an obstacle including the obstacle detection device that can accurately detect an obstacle such as a parked vehicle and can significantly reduce the amount of processing compared to conventional methods. An object is to provide a detection system.

The obstacle detection method according to the first aspect of the present invention is an obstacle detection method for detecting the presence of an obstacle by processing each captured image acquired by two imaging devices. Based on this, the edge point is detected, the edge group connecting the adjacent edge points is identified, the area surrounding the identified edge group is divided into small areas, and each small area including the edge point is divided among the divided small areas. A correlation value is calculated between the other captured image and a corresponding area corresponding to the small area is identified by the other captured image based on the calculated correlation value, and a parallax between the small area and the corresponding area is determined. Based on the candidate area, the candidate area is extracted from the identified candidate area, and the obstacle is detected based on the feature amount of the extracted obstacle candidate area. It is characterized by .

  An obstacle detection device according to a second invention is an obstacle detection device that detects the presence of an obstacle by processing the respective captured images acquired by the two imaging devices, and sets the pixel value of each pixel of one captured image. Based on detection means for detecting edge points, means for specifying edge groups obtained by combining adjacent edge points detected by the detection means, and a region surrounding the edge group specified by the means is divided into small areas. Dividing means, a means for calculating a correlation value with the other captured image for each small area including an edge point among the small areas divided by the dividing means, and a correlation value calculated by the means Based on the means for identifying the corresponding region corresponding to the small region with the other captured image and the parallax between the small region and the corresponding region, the candidate region is identified with a substantially equal distance from the imaging device to the imaging target Specific means to do Extraction means for extracting an obstacle candidate area from the candidate area specified by the specifying means, and means for detecting an obstacle based on the feature amount of the obstacle candidate area extracted by the extraction means. Features.

  The obstacle detection apparatus according to a third aspect of the present invention is the obstacle detection device according to the second aspect, wherein the parallax of the connected candidate areas connected based on the means for connecting the candidate areas specified by the specifying means and the parallax of each of the candidate areas And a means for determining whether or not the statistical value calculated by the means is smaller than a predetermined statistical threshold, and the extracting means is configured such that the statistical value is smaller than the statistical threshold. A region constituted by the determined one or a plurality of connection candidate regions is extracted as an obstacle candidate region.

  An obstacle detection system according to a fourth aspect of the present invention includes two imaging devices that image the front of the vehicle and the obstacle detection device according to any one of the aforementioned inventions, and two obtained by imaging with the imaging device. The captured image is processed by the obstacle detection device to detect the presence of an obstacle ahead of the vehicle.

  In the first invention, the second invention, and the fourth invention, the edge point is detected based on the pixel value of each pixel of one of the picked-up images picked up by the two image pickup devices. As an operator for edge point detection, for example, a Pruit wit operator can be used. An edge group in which adjacent edge points are combined among the detected edge points is specified. An area surrounding the identified edge group is divided into small areas (blocks). For example, adjacent edge points in the edge image are grouped, a rectangular area surrounding the edge group is specified, and then the specified area is divided into small areas. By simply joining adjacent edge points, objects that exist at different distances (for example, pedestrians and backgrounds) may be joined, but they are recombined using distance information after being divided into small areas. By doing so, objects existing at different distances can be separated and detected. Among the divided small areas (blocks), a small area including an edge point (hereinafter referred to as “edge area”) is used as a reference area, and the other captured image (reference image) is searched to calculate a correlation value. The region having the largest correlation value in the reference image is specified as the corresponding region corresponding to the edge region, and the parallax between the edge region and the corresponding region is calculated. Based on the parallax calculated for each edge region, one or a plurality of edge regions (candidate regions) having substantially the same distance from the imaging device (for example, an intermediate position between the two imaging devices) to the imaging target are specified.

An obstacle candidate area is extracted from the identified edge area (candidate area). The obstacle candidate area can be extracted by, for example, combining edge areas that are substantially equidistant. An obstacle is detected based on the extracted feature amount (for example, the size, shape, etc.) of the obstacle candidate area. As a result, the search process is not performed for the small area (block) that does not include the edge point, and the search process is performed using only the portion where the edge point exists, so that the processing amount can be reduced. In addition, since small areas that do not include edge points have few features and are likely to cause search errors, search errors can be suppressed by not performing search processing on these small areas. Furthermore, if there is a luminance difference between an obstacle such as a pedestrian or animal and the background, an edge point can be detected, and the luminance difference also occurs in summer or daytime, so the luminance of the obstacle such as a pedestrian or animal Is not necessary, and the obstacle can be detected with high accuracy even in summer or daytime.

  In the third invention and the fourth invention, the identified candidate areas are connected, and a statistical value (for example, an average value) of the parallax of the linked candidate areas is calculated based on the parallax of each candidate area. To do. It is determined whether or not the calculated statistical value is smaller than a predetermined second threshold, and an area composed of one or a plurality of connection candidate areas determined that the statistical value is smaller than the predetermined second threshold is an obstacle candidate Extract as a region. As a result, it is possible to extract only those having the same statistical parallax value (that is, the distance to the imaging target) from among the plurality of connection candidate areas, excluding connection candidate areas having different distances, and the like. Obstacle candidate regions composed of distance connection candidate regions can be accurately extracted.

  In the present invention, the search process is not performed for small areas (blocks) that do not include edge points, and the search process is performed using only the portion where the edge points exist, so that the processing amount can be reduced. In addition, since small areas that do not include edge points have few features and are likely to cause search errors, search errors can be suppressed by not performing search processing on these small areas.

  Hereinafter, the present invention will be described with reference to the drawings illustrating embodiments thereof. In the following embodiments, a case where a pedestrian existing in front of a vehicle is detected based on an image captured by an imaging device will be described as an example. The present invention can also be applied to the case where an obstacle such as a vehicle is detected, and an infrared imaging device can also be used.

  FIG. 1 is a schematic diagram showing a configuration of an obstacle detection system according to an embodiment of the present invention. In the present embodiment, the imaging devices 1 and 1 that image the front of the vehicle are juxtaposed in the vicinity of the room mirror in the vehicle interior. The imaging devices 1 and 1 are imaging devices that use visible light having a wavelength of 0.4 to 0.8 μm. The image data captured by the imaging devices 1 and 1 is transmitted to the obstacle detection device 3 connected via the video cable 7 corresponding to an analog video system such as NTSC or a digital video system. In addition, by imaging the common imaging region between the two imaging devices 1 and 1, the distance to the imaging target (that is, the parallax of the imaging target in both captured images) can be obtained based on the principle of triangulation.

  The obstacle detection device 3 is connected to the imaging device 1 and the display device 4 provided with an operation unit via the cable 8 corresponding to a video system such as NTSC, VGA, DVI, etc. An output device such as an alarm device 5 that emits an audible warning by a sound effect or the like is connected via an in-vehicle LAN cable 6 compliant with CAN.

  FIG. 2 is a block diagram illustrating a configuration of the imaging apparatus 1. The image capturing unit 11 includes an image sensor such as a charge coupled device (CCD) or a complementary metal oxide semiconductor (CMOS) that converts an optical signal into an electric signal in a matrix. The image capturing unit 11 reads a visible light image around the vehicle as an electric signal, and transmits the read electric signal to the signal processing unit 12.

The signal processing unit 12 is an LSI, converts an electrical signal received from the image capturing unit 11 into a digital signal, and stores the digital signal in the image memory 13 as image data. Note that it is not essential to temporarily store the image data in the image memory 13, and the image data may be transmitted directly to the obstacle detection device 3 via the video output unit 14.

  The video output unit 14 is an LSI, and outputs video data to the obstacle detection device 3 via a video cable 7 compatible with an analog video system such as NTSC or a digital video system.

  FIG. 3 is a block diagram showing the configuration of the obstacle detection apparatus 3. The obstacle detection device 3 includes an LSI 31, an image memory 32, a RAM 33, a video input unit 34, a video output unit 35, a communication interface unit 36, and the like.

  The video input unit 34 inputs video signals from the imaging devices 1 and 1. The video input unit 34 stores the image data input from the imaging devices 1 and 1 in the image memory 32 in synchronization with each frame. The video output unit 35 outputs image data to the display device 4 such as a liquid crystal display via the cable 8, and the communication interface unit 36 outputs a buzzer, a speaker, a display unit, etc. via the in-vehicle LAN cable 6. A signal for synthesized sound and data for character display are transmitted to the alarm device 5 provided. When a pedestrian is detected, the alarm device 5 can display, for example, a letter “DANGER” and emit a warning sound. In addition, the detected pedestrian image can be displayed on the display device 4.

  The image memory 32 is an SRAM, flash memory, SDRAM, or the like, and stores image data input from the imaging devices 1 and 1 via the video input unit 34.

  The LSI 31 reads out the image data stored in the image memory 32 in units of frames, and performs predetermined image processing on the read captured images (image data) in units of frames. For example, the LSI 31 detects an edge point in one captured image, an edge grouping process that groups adjacent edge points in an edge image composed of the detected edge points, and divides a rectangular region surrounding the edge group into blocks. Division processing, processing that searches a reference image (the other captured image) using a block including an edge point among the divided blocks to identify a corresponding region, and an equidistant block based on a parallax between the reference region and the corresponding region A process for extracting pedestrian candidate regions by grouping the pedestrians, a determination process for performing pedestrian determination based on feature amounts, and the like are performed. Details of processing in the LSI 31 will be described later.

  Next, the operation of the obstacle detection device 3 of the present invention will be described. An example in which a pedestrian is detected as an obstacle will be described below. FIG. 4 is a flowchart showing the procedure of the pedestrian detection process of the obstacle detection device 3.

  The LSI 31 reads out image data stored in the image memory 32 in units of frames (S11), and picks up one of the read out images in units of frames (an image obtained by imaging with one imaging device 1). Edge points are detected based on pixel values (for example, RGB values, luminance values, etc.) of each pixel of the image (S12).

  FIG. 5 is an example of an operator used for edge point detection. The operator shown in FIG. 5 is a Pruit wit operator and can mainly detect the vertical edge point of the captured image. The edge point detection operator is not limited to this, and other operators may be used.

  The LSI 31 groups adjacent edge points in the edge image obtained by detecting the edge points, and extracts an edge group (S13).

FIG. 6 is an explanatory diagram showing an example of an edge group. As shown in FIG. 6, the edge group is composed of adjacent edge points and is represented by a continuous straight line or curve. In FIG. 6, edge groups A1, A2, A3, A4, A5, etc. are extracted as an example.

  The LSI 31 sets a rectangular area circumscribing each edge group, and divides the set rectangular area into blocks (small areas) having a predetermined size (for example, 4 × 4 pixels) (S14). Further, the LSI 31 specifies a block (edge region) including an edge point among the divided blocks for calculating a correlation value described later. The block size is not limited to the above example.

  FIG. 7 is an explanatory diagram showing an example of divided blocks. Although FIG. 7 shows an example in which the edge group A1 is divided, the same applies to other edge groups. As shown in FIG. 7, edge group A1 is divided into 15 blocks of 3 × 5, and among the divided blocks, seven blocks including edge points (blocks with a pattern in FIG. 7) are identified. Has been.

  The LSI 31 uses the edge region as a reference region, the other captured image (the captured image obtained by imaging with the other imaging device 1) as a reference image, and the region (corresponding region) corresponding to the edge region in the reference image. Specify (S15). Specifically, a correlation calculation is performed by setting an area (reference area) having the same size as the edge area in the reference image, and the reference area giving the highest correlation value is specified as the corresponding area. The calculation of the correlation value is performed based on, for example, an expression represented by Expression 1.

  Here, N is the total number of pixels in the edge region and the reference region, k is an integer from 0 to N-1, Fk is the luminance value of the kth pixel in the edge region, and Gk is the kth pixel in the reference region. The luminance value and R represent a correlation value. The correlation value calculation method is not limited to this, and any method such as an absolute difference sum method or a normalized cross-correlation method may be used.

  The LSI 31 calculates the parallax between the edge region and the identified corresponding region (S16). The parallax can be calculated based on the coordinates of the edge region or the corresponding region in each captured image.

  FIG. 8 is an explanatory diagram illustrating an example of calculating parallax for each edge region. As shown in FIG. 8, the parallax (a numerical value in parentheses) with the corresponding region corresponding to the edge region is calculated. The parallax for each edge region from the upper right to the lower left is “5”, “9”, “11”, “10”, “9”, “10”, “11”.

The LSI 31 compares the parallax difference (difference) between the edge areas with a first threshold (for example, “3”), groups edge areas where the parallax difference (difference) is smaller than the first threshold, Edge groups are re-extracted (S17). As a result, each of the edge regions with parallax “9”, “11”, “10”, “9”, “10”, “11” has a difference in parallax from the other edge regions is smaller than the first threshold value. Although grouped, an edge region having a parallax of “5” is excluded because a difference in parallax from other edge regions is larger than the first threshold value. Here, the fact that the difference in parallax is smaller than the first threshold means that the distance to the same imaging target imaged by the two imaging devices 1 and 1, more specifically, the optical axes of the two imaging devices 1 and 1. That is, the distance from the intermediate position to the imaging target is substantially equal.

  FIG. 9 is an explanatory diagram illustrating an example of edge groups re-extracted based on parallax. As shown in FIG. 9, in the edge group A1, the edge regions (blocks indicated by broken lines in the figure) having a parallax of “5” are excluded because they exist at different distances, and a new edge group S1 is extracted. ing. In this case, a configuration in which a block that does not include an edge point is excluded and a new edge group S1 is extracted may be employed. As a result, it is possible to extract only those having substantially the same parallax (that is, the distance to the imaging target) from a plurality of blocks, and when a part of the imaging targets having different distances are overlapped and captured. Even in such a case, it is possible to accurately specify only edge groups that are substantially equidistant by excluding imaging targets having different distances.

  The LSI 31 calculates the parallax of the newly re-extracted edge group S1 based on the average value obtained by averaging the parallax of each edge region (S18). Instead of the average value, other statistical values such as the mode value and the intermediate value may be calculated. Also, the LSI 31 re-extracts the edge groups S2, S3,... By performing the above-described processing for the other edge groups A2, A3,.

  Since the parallaxes of the edge regions grouped in FIG. 8 are “9”, “11”, “10”, “9”, “10”, and “11”, respectively, as shown in FIG. The parallax of the edge group S1 is “10”, which is the average value of the parallaxes.

  FIG. 10 is an explanatory diagram showing examples of all extracted edge groups. As shown in the figure, edge groups S1, S2,..., S10 are extracted, and the parallax of each edge group is a value indicated by a numerical value in parentheses.

  The LSI 31 compares the parallax between the re-extracted edge groups, groups the edge groups whose parallax difference is equal to or smaller than a second threshold (for example, “4”), and extracts the pedestrian candidate areas (S19). At this time, since there is a correlation between the size of the pedestrian in the captured image and the parallax (distance to the pedestrian), the edge groups that become inappropriate as the size of the pedestrian when grouped are Extract as separate pedestrian candidate areas without grouping.

  FIG. 11 is an explanatory diagram illustrating an example of extraction of pedestrian candidate regions. Among the edge groups extracted in FIG. 10, the parallaxes of the edge groups S1, S2, S3, S4, and S5 are “10”, “11”, “11”, “10”, and “9”, respectively. The edge groups are grouped and the pedestrian candidate area V1 is extracted. Further, the parallaxes of the edge groups S6, S7, S8, and S9 are “10”, “10”, “11”, and “12”, respectively, and these edge groups are grouped to form a pedestrian candidate region V2. Has been extracted.

The parallax of the edge groups S1, S2, S3, S4, and S5 and the parallax of the edge groups S6, S7, S8, and S9 have a small difference in parallax, but the pedestrian candidate area when both are grouped is a pedestrian Since the size is inappropriate, it is extracted as a separate pedestrian candidate area. Further, the parallax of the edge group S10 is “16”, and since the difference from the parallax of the other edge groups is large, it is extracted as a separate pedestrian candidate area as an imaging target having a different distance. In addition, when extracting a pedestrian candidate area | region, you may extract the area | region containing an edge group as a pedestrian candidate area | region, and without including all edge groups,
Depending on the shape of the entire edge group, an area composed of a predetermined contour can be extracted as a pedestrian candidate area.

  The LSI 31 extracts a feature amount (for example, shape, size, etc.) of the extracted pedestrian candidate area, performs pedestrian determination based on the extracted feature amount (S20), and determines a feature amount suitable for a pedestrian. The pedestrian candidate area that is held is detected as a pedestrian. Note that a learning algorithm such as support vector machine (SVM) or boosting can be used for the pedestrian determination process. When detecting a pedestrian, the LSI 31 can highlight the pedestrian detected on the display device 4 by surrounding it with a rectangular frame, for example. Alternatively, the alarm device 5 is controlled to display a warning character and generate a warning sound. The extracted edge group, the specified edge region, and the extracted pedestrian candidate region are not limited to a plurality, and may be one.

  As described above, in the present invention, search processing is not performed for blocks that do not include edge points, and search processing is performed using only blocks that have edge points, so that the amount of processing can be reduced. it can. In addition, since small areas that do not include edge points have few features and are likely to cause search errors, search errors can be suppressed by not performing search processing on these small areas. In addition, if there is a luminance difference between an obstacle such as a pedestrian or animal and the background, an edge point can be detected, and the luminance difference also occurs in summer or daytime, so the luminance of the obstacle such as a pedestrian or animal Is not necessary, and the obstacle can be detected with high accuracy even in summer or daytime. In addition, even when a part of imaging targets with different distances are overlapped and imaged, it is possible to separate imaging targets with different distances and accurately specify only substantially equidistant blocks, Among edge groups in which a plurality of blocks are grouped, edge groups with different distances can be separated, and obstacle candidate regions (pedestrian candidate regions) composed of substantially equal distance edge groups can be accurately extracted. .

  In the above-described embodiment, an example in the case of detecting a pedestrian has been described. However, the obstacle to be detected is not limited to a pedestrian, but is an obstacle such as an animal, another vehicle, or a foreign object on a road. Even if it exists, this invention is applicable.

  In the above embodiment, the visible light imaging device is used. However, the imaging device is not limited to this, and a configuration using a near infrared or far infrared imaging device may be used. A near-infrared imaging device acquires near-infrared light having a wavelength of 0.8 to 3 μm, and a far-infrared imaging device acquires far-infrared light having a wavelength of 8 to 12 μm. Further, the installation location of the imaging device is not limited to the vicinity of the rearview mirror, and can be installed in a bumper, a front grill, or the like.

It is a mimetic diagram showing composition of an obstacle detection system concerning an embodiment of the invention. It is a block diagram which shows the structure of an imaging device. It is a block diagram which shows the structure of an obstruction detection apparatus. It is a flowchart which shows the procedure of the pedestrian detection process of an obstruction detection apparatus. It is an example of an operator used for edge point detection. It is explanatory drawing which shows an example of an edge group. It is explanatory drawing which shows the example of the divided | segmented block. It is explanatory drawing which shows the example of calculation of the parallax for every edge area | region. It is explanatory drawing which shows the example of the edge group re-extracted based on parallax. It is explanatory drawing which shows the example of all the extracted edge groups. It is explanatory drawing which shows the example of extraction of a pedestrian candidate area | region.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Imaging device 3 Obstacle detection device 4 Display device 5 Alarm device 31 LSI
32 Image memory 33 RAM
34 Video Input Unit 35 Video Output Unit 36 Communication Interface Unit

Claims (4)

In the obstacle detection method for detecting the presence of an obstacle by processing each captured image acquired by two imaging devices,
An edge point is detected based on the pixel value of each pixel of one captured image,
Identify edge groups that combine adjacent edge points,
Divide the area surrounding the identified edge group into small areas,
Of the divided small areas, calculate a correlation value with the other captured image for each small area including an edge point,
Based on the calculated correlation value, the corresponding area corresponding to the small area is specified in the other captured image,
Based on the parallax between the small area and the corresponding area, identify candidate areas having substantially the same distance from the imaging device to the imaging target;
Extract obstacle candidate areas from the identified candidate areas,
An obstacle detection method characterized by detecting an obstacle based on a feature amount of an extracted obstacle candidate region. In the obstacle detection device that detects the presence of an obstacle by processing each captured image acquired by two imaging devices,
Detection means for detecting an edge point based on the pixel value of each pixel of one captured image;
Means for specifying an edge group obtained by combining adjacent edge points detected by the detection means;
Dividing means for dividing an area surrounding the edge group specified by the means into small areas;
Means for calculating a correlation value with the other captured image for each small region including an edge point among the small regions divided by the dividing unit;
Means for identifying a corresponding area corresponding to the small area in the other captured image based on the correlation value calculated by the means;
Identifying means for identifying candidate areas having substantially the same distance from the imaging device to the imaging target based on the parallax between the small area and the corresponding area;
Extracting means for extracting an obstacle candidate area from the candidate area specified by the specifying means;
An obstacle detection apparatus comprising: means for detecting an obstacle based on the feature amount of the obstacle candidate area extracted by the extraction means. Means for connecting the candidate regions identified by the identifying means;
Means for calculating a statistical value of the parallax of linked candidate regions based on the parallax of each of the candidate regions;
Means for determining whether or not the statistical value calculated by the means is smaller than a predetermined statistical threshold,
The extraction means includes
The region configured by one or a plurality of connection candidate regions determined to have the statistical value smaller than the statistical threshold value is configured to be extracted as an obstacle candidate region. Obstacle detection device.   Two imaging devices that image the front of the vehicle and the obstacle detection device according to claim 2 or 3 are provided, and two captured images obtained by imaging with the imaging device are captured by the obstacle detection device. An obstacle detection system characterized by processing to detect the presence of an obstacle ahead of the vehicle.

Images