JP2010092429A - Vehicle surrounding monitoring device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a vehicle surrounding monitoring device capable of recognizing whether or not an object is a four-legged animal with high accuracy. <P>SOLUTION: By the vehicle surrounding monitoring device, it is determined that the object corresponds to the four-legged animal on condition that closed elements and open elements are extracted in different time from an object area as an area where the object extracted from a pick up image exists. In addition, it is determined that the object corresponds to the four-legged animal on condition that the closed elements are intermittently extracted from the object area. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a vehicle periphery monitoring device that monitors the periphery of a vehicle using a captured image obtained by an imaging unit mounted on the vehicle.

A region corresponding to the pedestrian's head, which is an object, is extracted from the infrared image captured by the imaging device, and the region corresponding to the pedestrian's torso is extracted based on the head region. A technique for recognizing the presence of a pedestrian and informing the driver of the vehicle of the presence of the pedestrian is known (see Patent Document 1).
Japanese Patent Laid-Open No. 11-328364

  However, there are cases where not only pedestrians (human beings) but also quadruped animals such as deer exist around the vehicle. In this case, it is difficult to determine whether or not the object is a quadruped animal even if a method for determining whether or not the object is a human is employed. For this reason, it becomes difficult to inform the driver of the vehicle of the presence of a quadruped animal, and to drive the vehicle so that the driver avoids contact with the object.

  Then, an object of this invention is to provide the vehicle periphery monitoring apparatus which can recognize with high precision whether a target object is a quadruped animal.

  A vehicle periphery monitoring device according to a first aspect of the present invention is a vehicle periphery monitoring device that monitors the periphery of a vehicle using a captured image obtained by an imaging unit mounted on the vehicle, and includes a region where an object exists from the captured image. A target area extracting means for extracting as a target area; a closed element composed of a high luminance pixel group surrounding a low luminance pixel group from the target area extracted by the target area extraction means; An object classification means for determining that the object corresponds to the quadruped animal is provided on the condition that an open element having a shape in which a lower part of the element is cleaved is extracted at a different time. .

  According to the vehicle periphery monitoring device of the first aspect of the invention, it is extracted from a captured image (in addition to the original image captured by the imaging means, an image obtained by processing the original image is included). On the condition that the closed element and the open element are extracted at different times from the object area as the area where the target object exists, it is determined that the object corresponds to a quadruped animal. The “quadruped animal” does not mean all the quadruped animals, but means a specific quadruped animal that can be classified by the method of the present invention. When a quadruped animal that is walking or running is observed from the side, it is highly likely that it is confirmed that the intersection and separation of the front and rear leg intersections or the ground contact part are repeated. In addition, the intersection part or the ground contact part is located below the base of the leg. In view of these points, it is possible to recognize with high accuracy whether or not the object represented by the high luminance region extracted from the captured image by the determination method as described above is a quadruped.

  A vehicle periphery monitoring device according to a second aspect of the present invention is a vehicle periphery monitoring device that monitors the periphery of a vehicle using a captured image obtained by an imaging unit mounted on the vehicle, and includes a region where an object is present from the captured image. Object area extraction means for extracting as an object area, and closed elements composed of high-luminance pixel groups surrounding low-luminance pixel groups intermittently from the object area extracted by the object area extraction means An object classification means for determining that the object corresponds to a quadruped animal is provided on the condition that it is extracted.

  According to the vehicle periphery monitoring device of the second aspect of the present invention, it is required that the closed element is intermittently extracted from the object area as the area where the object extracted from the captured image is present. It is determined to be a paw animal. The closure element is likely to correspond to at least a portion of the lateral quadruped animal's torso and the front and rear legs that intersect from the side or are close to the ground contact area. When a walking or running quadruped animal is observed from the side, it is likely that a pair of front and rear legs that are crossing or in close proximity to the ground contact area will be confirmed intermittently. In view of these points, it is possible to recognize with high accuracy whether or not the object represented by the high luminance region extracted from the captured image by the determination method as described above is a quadruped.

  An embodiment of a vehicle periphery monitoring device of the present invention will be described. First, the structure of the vehicle periphery monitoring apparatus of this embodiment is demonstrated. The vehicle periphery monitoring device includes an image processing unit 1 shown in FIGS. 1 and 2. The image processing unit 1 is connected to two infrared cameras 2R and 2L as imaging means for capturing an image in front of the vehicle 10 and detects the yaw rate of the vehicle 10 as a sensor for detecting the traveling state of the vehicle 10. The yaw rate sensor 3 that detects the vehicle 10, the vehicle speed sensor 4 that detects the traveling speed (vehicle speed) of the vehicle 10, and the brake sensor 5 that detects whether the vehicle 10 is braked or not are connected. Further, the image processing unit 1 includes a speaker 6 for outputting auditory notification information such as voice, and a display device for displaying captured images and visual notification information captured by the infrared cameras 2R and 2L. 7 is connected. The infrared cameras 2R and 2L correspond to the image pickup means in the present invention. When the distance to the object is measured by a radar, only one infrared camera may be mounted on the vehicle 10.

  The image processing unit 1 includes an ECU (electronic control unit) that includes an A / D conversion circuit, a memory such as a CPU, a RAM, and a ROM, an I / O circuit, and a bus that connects these. Analog signals output from the infrared cameras 2R, 2L, the yaw rate sensor 3, the vehicle speed sensor 4, and the brake sensor 5 are converted into digital data by the A / D conversion circuit, and based on the data digitized by the CPU, humans (pedestrians) When a detected object satisfies a predetermined notification requirement, a process of issuing a notification to the driver by the speaker 6 or the display device 7 is executed. The image processing unit 1 has functions as an object extraction unit and an object classification unit.

  As shown in FIG. 2, the infrared cameras 2 </ b> R and 2 </ b> L are attached to the front portion of the vehicle 10 (the front grill portion in the figure) in order to image the front of the vehicle 10. In this case, the infrared cameras 2R and 2L are respectively arranged at a position on the right side and a position on the left side of the center of the vehicle 10 in the vehicle width direction. These positions are symmetrical with respect to the center of the vehicle 10 in the vehicle width direction. The infrared cameras 2R and 2L are fixed so that their optical axes extend in parallel in the front-rear direction of the vehicle 10 and the heights of the respective optical axes from the road surface are equal to each other. The infrared cameras 2R and 2L have sensitivity in the far-infrared region, and have a characteristic that the higher the temperature of an object to be imaged, the higher the level of the output video signal (the higher the luminance of the video signal). is doing.

  In the present embodiment, the display device 7 includes a head-up display 7a (hereinafter referred to as HUD 7a) that displays image information on the front window of the vehicle 10. As the display device 7, a display provided integrally with a meter that displays a traveling state such as the vehicle speed of the vehicle 10 or a display provided in an in-vehicle navigation device is used instead of or together with the HUD 7 a. May be.

  Next, basic functions of the vehicle periphery monitoring apparatus having the above-described configuration will be described with reference to the flowchart of FIG. The basic processing contents of the flowchart of FIG. 3 are the same as the processing contents described in FIG. 3 of Japanese Patent Application Laid-Open No. 2001-6096 and FIG. 3 of Japanese Patent Application Laid-Open No. 2007-310705, for example.

  Specifically, first, the infrared image signals of the infrared cameras 2R and 2L are input to the image processing unit 1 (FIG. 3 / STEP 11), and then a series of processes are executed by the image processing unit 1. That is, the output signals of the infrared cameras 2R and 2L are A / D converted (FIG. 3 / STEP 12), a gray scale image is generated from the A / D converted infrared image (FIG. 3 / STEP 13), and the reference gray scale is obtained. The image (right image) is binarized (FIG. 3 / STEP 14). In addition, an area where the object exists is extracted as the object area S from the binarized image. Specifically, the pixel group constituting the high-luminance area of the binarized image is converted into run-length data (FIG. 3 / STEP 15), and a label (identifier) is assigned to each line group that overlaps in the vertical direction of the reference image. Is attached (FIG. 3 / STEP 16), and each group of lines is extracted as an object (FIG. 3 / STEP 17). As a result, a high-luminance region constituted by a group of high-luminance pixels (pixel value = 1 pixel) as indicated by the oblique lines in FIG. 6A is extracted as an object. Furthermore, an area defined by a line surrounding the entire object indicated by hatching in FIG. 6A is set as the object area A.

  Subsequently, the position of the center of gravity of each object (position on the reference image), the area, and the aspect ratio of the circumscribed rectangle are calculated (FIG. 3 / STEP 18). Further, the tracking of the target object is performed, and the identity of the target object is determined every calculation processing cycle of the image processing unit 1 (FIG. 3 / STEP 19). Further, the outputs (vehicle speed detection value and yaw rate detection value) of the vehicle speed sensor 4 and the yaw rate sensor 5 are read (FIG. 3 / STEP 20).

  Further, in parallel with the calculation of the aspect ratio of the circumscribed rectangle and the tracking of the object during the time, an area corresponding to each object (for example, the area of the circumscribed rectangle of the object) in the reference image is extracted as a search image. (FIG. 3 / STEP 31). Further, by executing the correlation calculation, an image (corresponding image) corresponding to the search image is searched and extracted from the left image (FIG. 3 / STEP 32). Further, the real space distance from the vehicle 10 to the object (the distance in the front-rear direction of the vehicle 10) is calculated (FIG. 3 / STEP 33).

  Further, the real space position (relative position with respect to the vehicle 10) of each object is calculated (FIG. 3 / STEP 21). Further, the X component of the real space position (X, Y, Z) of the object (see FIG. 2) is corrected according to the time-series data of the turning angle (see STEP 20) (FIG. 3 / STEP 22). Further, the relative movement vector of the object with respect to the vehicle 10 is estimated (FIG. 3 / STEP 23), and the driver's attention when it is determined that there is a possibility of contact between the vehicle 10 and the object based on the relative movement vector. A notification determination process for issuing a notification for calling out is executed (FIG. 3 / STEP 24). When it is determined that the object satisfies the notification determination requirement (FIG. 3 / STEP 24... YES), a notification output determination process is performed to determine whether or not notification is necessary for the object (FIG. 3 / STEP 25). If it is determined that notification is necessary (FIG. 3 / STEP 25... YES), the driver is notified that the possibility of contact with the vehicle 10 is high (FIG. 3 / STEP 26). Specifically, the situation is notified by outputting a sound through the speaker 6 or by highlighting an object with a color, a frame, or the like on the HUD 7a. The above is the overall operation of the vehicle periphery monitoring device of the present embodiment. In addition, the structure which performs the process of STEP11-18 by the image processing unit 1 is equivalent to the target object area | region extraction means of this invention.

  Next, the notification determination process in STEP 24 of the flowchart of FIG. 3 will be described using the flowchart of FIG. The basic processing content of the flowchart of FIG. 4 is the same as the processing content described in FIG. 4 of Japanese Patent Application Laid-Open No. 2001-6096 and FIG.

  Specifically, a contact determination process is first executed (FIG. 4 / STEP 41). In the contact determination process, when the distance between the vehicle 10 and the object is equal to or less than a value obtained by multiplying the relative speed Vs and the margin time T, it is determined that there is a possibility of contact. If it is determined that there is a possibility that the vehicle 10 and the object are in contact with each other within the margin time T (FIG. 4 / STEP 41... YES), it is determined whether or not the object exists in the approach determination area. (FIG. 4 / STEP 42). When it is determined that the object does not exist in the approach determination area (FIG. 4 / STEP 42... NO), it is determined whether or not the object may enter the approach determination area and come into contact with the vehicle 10. (FIG. 4 / STEP 43). Specifically, there is a possibility of contact when the object exists in the approach determination area outside the approach determination area and the movement vector of the object calculated in STEP 23 is toward the approach determination area. It is determined that there is.

  On the other hand, when it is determined that the object is present in the approach determination area (FIG. 4 / STEP 42... YES), or when it is determined that there is a possibility of contact between the vehicle 10 and the object (FIG. 4 / (STEP 43... YES), an artificial structure determination process for determining whether or not the object is an artificial structure is executed (FIG. 4 / STEP 44). Thereby, for example, when a feature on the outer shape that is impossible for a pedestrian is detected from an object, it is determined that the object corresponds to an artificial structure. When it is determined that the object does not correspond to an artificial structure (FIG. 4 / STEP 44... NO), it is determined whether the object corresponds to a pedestrian (FIG. 4 / STEP 45). Specifically, it is determined whether or not the object corresponds to a pedestrian from the characteristics such as the shape and size of the image area of the object on the gray scale image and luminance dispersion. Then, when it is determined that the object corresponds to a pedestrian (FIG. 4 / STEP 45... YES), the detected object is determined as a notification or alerting target (FIG. 4 / STEP 46).

  When it is determined that the object is not a pedestrian (FIG. 4 / STEP 45... NO), it is determined whether the object corresponds to a quadruped animal (FIG. 4 / STEP 47). When it is determined that the target object corresponds to a quadruped animal (FIG. 4 / STEP 47... YES), the image processing unit 1 determines that the target object is a notification target (FIG. 4 / STEP 46). When it is determined that there is no possibility of contact between the vehicle 10 and the object (FIG. 4 / STEP 43... NO), when the object is determined to be an artificial structure (FIG. 4 / STEP 44... YES), or When it is determined that the object does not correspond to a quadruped animal (FIG. 4 / STEP 47... NO), this object is excluded from the notification target (FIG. 4 / STEP 48).

  Next, an animal determination process that is a main function of the vehicle periphery monitoring device of the present invention will be described with reference to the flowchart of FIG.

  First, in the object area A, a maximum area B constituted by a high luminance pixel group whose luminance is included in the highest luminance range is extracted, and the ratio of the horizontal width W to the vertical width (height) H of the area B is extracted. Is calculated as the object aspect ratio (= W / H) (FIG. 5 / STEP 102). Thereby, for example, an area defined by a broken line surrounding a part of the target object (high luminance image area) indicated by diagonal lines in FIG. 6A is extracted as the area B, and the aspect ratio W / H is calculated. Is done. Then, it is determined whether or not the aspect ratio W / H is a reference value (for example, 1) or more (FIG. 5 / STEP 104). The calculation of the aspect ratio W / H and the comparison process with the reference value may be omitted.

  When it is determined that the aspect ratio W / H is less than the reference value (FIG. 5 / STEP 104... NO), it is determined that the object does not correspond to a quadruped animal (FIG. 5 / STEP 128). This determination is based on the idea that when the object is a quadruped animal, it is long and is likely to be extracted from a high-luminance region corresponding to a torso that is hotter than other body parts. .

  On the other hand, when it is determined that the aspect ratio W / H is equal to or greater than the reference value (FIG. 5 / STEP 104... YES), in the object area A, a designation composed of a high luminance area surrounding a group of low luminance areas. The presence / absence of a closed element of the shape is determined (FIG. 5 / STEP 106). Specifically, the object area A is scanned in the right direction in order from the lower end to the upper end of the object area A, so that the pixel value is “1 (represents a pixel in which the object exists)”. After changing to “0 (represents a pixel in which no object exists)”, a horizontal line segment constituted by horizontally arranged pixels that again becomes “1” is searched. As a result, for example, for the object indicated by hatching in FIG. 7A, from the negative edge point (● in the figure) where the pixel value changes from “1” to “0” in the right direction of the image, A line segment extending from a luminance pixel or a plurality of continuous low-luminance pixel groups to a positive edge point (◯ in the figure) where the pixel value changes from “0” to “1” is searched for as a horizontal line segment. A horizontal line segment whose length is less than the threshold value can be removed as noise.

  Further, the first determination flag f1 for the horizontal line located locally at the bottom is set, and the second determination flag f2 for the horizontal line located locally at the top is set (FIG. 7A). reference). The horizontal line segment located locally at the bottom means a horizontal line segment that does not overlap when it is shifted downward by one pixel. The horizontal line segment located locally at the top means a horizontal line segment in which there is no overlapping horizontal line segment when shifted upward by one pixel. Specifically, when the horizontal line segment located locally at the bottom is shifted by one pixel below the image, all the pixels constituting the horizontal line segment or all but one or two pixels are high-luminance pixels. , The first determination flag f1 for the horizontal line is set to “1”. In addition, when the horizontal line segment located locally at the top is shifted by one pixel above the image, all pixels constituting the horizontal line segment or all pixels except for one or two pixels overlap with the high luminance pixel. The second determination flag f2 for the horizontal line is set to “2”. In other cases, the determination flag f of the horizontal line located locally at the bottom or top is set to “0”.

  Subsequently, both the first determination flag f1 and the second determination flag f2 are attached to each horizontal line segment (see FIG. 7B). More specifically, the first determination flag f1 for the horizontal line that is immediately above the horizontal line segment for which the first determination flag f1 is 1 (up by one pixel) and overlaps the horizontal line segment in the vertical direction of the image is “ 1 ”. In addition, the second determination flag f2 is immediately below the horizontal line segment that is 2 (down by one pixel), and the second determination flag f2 of the horizontal line that overlaps the horizontal line segment in the vertical direction of the image is “2”. Is set. On the other hand, the first determination flag f1 of the horizontal line that is immediately above the horizontal line segment for which the first determination flag f1 is 0 and overlaps the horizontal line segment in the image vertical direction is set to “0”. Further, the second determination flag f2 that is immediately below the horizontal line segment for which the second determination flag f2 is 0 and that overlaps the horizontal line segment in the image vertical direction is set to “0”. Thereby, all the first determination flags f1 and the second determination flags f2 of the horizontal line segment are set.

  According to the above-described method, the total determination flag Σf = f1 + f2 for each horizontal line bridged by a portion where the top and bottom are both open (cut portion) in the high luminance region representing the object is “0”. The total determination flag Σf for each horizontal line that is bridged to a portion where the lower portion of the high-intensity region representing the object is open and the upper portion is closed is “1” (see FIG. 7B). On the other hand, the total determination flag Σf for each horizontal line bridged to a portion where the upper portion of the high luminance region representing the object is open and the lower portion is closed is “2”. The total determination flag Σf for each horizontal line that is bridged to the portion where the top and bottom are both closed in the high luminance region representing the object is “3”. When the total determination flag Σf is 3 and there are a plurality of horizontal line segments that are continuously overlapped in the vertical direction, it is determined that a “closed element” exists.

  In order to reduce the load and increase the speed of image processing, it may be determined whether or not there is a closed element in the local region S that is a part of the object region A. For example, as shown in FIG. 6B, a region defined by a thick line surrounding a high luminance region below the object region A can be set as the local region S. The horizontal width WS of the local area S is set to c1 · WA (0 <c1 <1) based on the horizontal width WA of the object area A, and the vertical width HS of the local area S is set based on the vertical width HA of the object area A. c2 · HA (0 <c2 <1) may be set. The coefficients c1 and c2 can be set arbitrarily. The lower end of the local region S may be aligned with or shifted from the lower end of the object region A, and the lateral end of the local region S may be aligned with or aligned with the lateral end of the object region A.

  Further, in order to determine whether or not the closed element has a designated shape, the vertical width LH and the horizontal width LW of the low-luminance region surrounded by the closed element are determined (see FIG. 7C). Specifically, among the plurality of horizontal line segments that have a total determination flag Σf of 3 and overlap in the vertical direction of the image, the longest line segment, that is, the horizontal line segment with the largest number of pixels constituting the horizontal line segment. Is determined as the horizontal width LW of the low luminance region surrounded by the closed elements. Further, from a negative edge point where the pixel value changes from “1” to “0” in the upward direction of the image through the midpoint (center pixel) of the longest horizontal line segment and the points (pixels) adjacent to both the left and right sides thereof, A line segment extending to a positive edge point where the pixel value changes from “0” to “1” through one low luminance pixel or a plurality of continuous low luminance pixel groups is searched for as a vertical line segment. In addition, the length of the longest vertical line segment among the vertical line segments is determined as the vertical width LH of the low luminance region surrounded by the closed elements.

  The ratio LH / LW of the vertical width LH to the horizontal width LW of the low-luminance region is within a specified ratio range [r1, r2] (for example, a range including “1” such as r1 = 0.5, r2 = 1.5, etc.). It is determined whether or not the shape of the closed element is a designated shape depending on whether or not it is included. Note that the shape determination of the closed element may be omitted.

  If it is determined that there is a closed element of the specified shape (FIG. 5 / STEP 106... YES), the initial value of the flag F of the object is set to “1” (FIG. 5 / STEP 108), and the image is displayed on the initial value. Timing from T = 0 by a timer (not shown) of the processing unit 1 is started (FIG. 5 / STEP 112). On the other hand, if it is determined that there is no closed element of the specified shape (FIG. 5 / STEP 106... NO), the initial value of the flag F is set to “0” (FIG. 5 / STEP 110), and then T = Timing from 0 is started (FIG. 5 / STEP 112).

  Subsequently, the presence / absence of a closed element having a specified shape in the object region A is determined according to the above-described method (FIG. 5 / STEP 114). If it is determined that there is a closed element of the designated shape (FIG. 5 / STEP 114... YES), the flag F of the object is maintained or changed to “1” (FIG. 5 / STEP 116). On the other hand, if it is determined that there is no closed element of the designated shape (FIG. 5 / STEP 114... NO), the flag F of the object is changed or maintained to “0” (FIG. 5 / STEP 118). In the object area A, the horizontal line segment having the total determination flag Σf of “3” is present in the horizontal line segment having the total determination flag Σf of “1”, that is, the lower part of the high luminance area representing the object. On the other hand, the flag F may be changed from “1” to “0” on the condition that a cross-linked horizontal line group appears in a portion where the top is closed (open element).

  Next, it is determined whether or not the timer measurement time T is less than the specified time T0 (FIG. 5 / STEP 120). When it is determined that the measurement time T is less than the specified time T0 (FIG. 5 / STEP 120... YES), it is further determined whether or not the number of changes of the flag F is a predetermined number (for example, 2) or more ( FIG. 5 / STEP 122). The designated time T0 is set based on a standard walking or running cycle of a quadruped animal. When the initial value of the flag F is “0”, the designated time T0 may be set longer than when the initial value of the flag F is “1”. The predetermined number of times may be the same or different between the case where the initial value of the flag F is “0” and the case where the initial value of the flag F is “1”. If it is determined that the number of changes of the flag F is equal to or greater than the predetermined number of times until the measurement time T reaches the specified time T0 (FIG. 5 / STEP 122... YES), the timer count is stopped (FIG. 5 / STEP 124). It is determined that the object corresponds to a quadruped animal (FIG. 5 / STEP 126). For example, as shown in FIG. 8 (a), until the time T measured by the timer reaches the specified time T0, the flag F = 1 in which the closed element exists, the open element instead of the closed element. When the flag F = 0 state and the flag F = 1 state appearing in order, the number of changes of the flag F reaches “2” as the predetermined number of times, the target object becomes a quadruped animal. It can be determined that this is the case. Further, as shown in FIG. 8B, the flag F = 0 state, the flag F = 1 state, and the flag F = 0 state until the time T measured by the timer reaches the specified time T0. By appearing in order, when the number of changes of the flag F reaches “2” as the predetermined number, it can be determined that the target object corresponds to a quadruped animal.

  Then, as described above, a report that alerts the driver of the vehicle 10 to this object can be executed (see FIG. 3 / STEP 26, FIG. 4 / STEP 46). On the other hand, if the measured time T of the timer reaches the specified time T0 without the number of changes of the flag F reaching the predetermined number (FIG. 5 / STEP 120... NO), it is determined that the object does not correspond to a quadruped animal. (FIG. 5 / STEP128).

  According to the vehicle periphery monitoring device that performs the function, it is required that the closed element and the open element are extracted at different times from the object area A as the area where the object extracted from the captured image exists. It is determined that the object corresponds to a quadruped animal (see FIG. 5 / STEP 122, STEP 126, FIGS. 8A and 8B). When a quadruped animal that is walking or running is observed from the side, it is highly likely that it is confirmed that the intersection and separation of the front and rear leg intersections or the ground contact part are repeated. In addition, the intersection part or the ground contact part is located below the base of the leg. In view of these points, it is possible to recognize with high accuracy whether or not the object represented by the high luminance region extracted from the captured image by the determination method as described above is a quadruped.

  Further, it is determined that the closed object is intermittently extracted from the object area A, and it is determined that the object corresponds to a quadruped animal (see FIG. 5 / STEP122, STEP126, and FIG. 8A). . The closure element is likely to correspond to at least a portion of the lateral quadruped animal's torso and the front and rear legs that intersect from the side or are close to the ground contact area. When a walking or running quadruped animal is observed from the side, it is likely that a pair of front and rear legs that are crossing or in close proximity to the ground contact area will be confirmed intermittently. In view of these points, it is possible to recognize with high accuracy whether or not the object represented by the high luminance region extracted from the captured image by the determination method as described above is a quadruped.

  Furthermore, on the condition that the shape of the low luminance pixel group or the low luminance region surrounded by the closed element corresponds to the specified shape, it is determined that the object from which the closed element is extracted corresponds to a quadruped animal (FIG. 5 / STEP 106, STEP 114, STEP 122, see FIG. 7C). Accordingly, when a lateral quadruped animal is observed in a state where the front legs and the rear legs cross each other or the ground contact parts of the front legs and the rear legs are close to each other, the front legs and the rear legs In view of the standard shape of the region surrounded by the hind legs, it is even higher whether the object represented by the high luminance region extracted from the captured image by the determination method as described above is a quadruped animal. Can be recognized with accuracy.

  In the embodiment, the horizontal line segment is searched by scanning the object region A in the horizontal direction of the image, and the presence or absence of the closed element is determined based on the evaluation result of the total determination flag Σf of the horizontal line segment. Vertical line segment by scanning the object area A in the vertical direction of the image (pixel line segment extending from the negative edge point to the positive edge point through one low-luminance pixel or a plurality of low-luminance groups that are vertically continuous) May be searched, and the presence or absence of a closed element may be determined based on the evaluation result of the total determination flag for the vertical line segment. Further, depending on whether or not the correlation value between a part of the high brightness area in the object area A and the template stored or stored in advance in the memory exceeds the reference value, the object area The presence or absence of the A closed element may be determined. As various templates, templates of a plurality of patterns in which some or all of the shapes, sizes, and arrangements (intervals) of the reference elements are different may be prepared. The template may be created based on an image of a specific kind of quadruped animal such as a sideways dog, deer, or horse imaged by the infrared cameras 2R and 2L. The presence / absence of a closed element may be determined based on the shape of a thinned image obtained by thinning the high-intensity region extracted from the object region A.

  In the embodiment, whether or not the shape of the low-luminance region or the low-luminance pixel group surrounded by the closed element corresponds to the specified shape is determined based on the vertical width LH and the horizontal width LW. The shape of the low luminance pixel group may be determined based on the change pattern of the length of the horizontal line segment. For example, it is determined that the low-luminance region has a specified shape that continuously or intermittently widens as it goes from the lower end to the upper end, and then continuously or intermittently becomes narrower as it goes to the upper end. May be.

Configuration explanatory diagram of the vehicle periphery monitoring device of the present invention Explanatory drawing about a vehicle equipped with a vehicle periphery monitoring device Flow chart showing functions of image processing unit constituting vehicle periphery monitoring device Flow chart showing functions of image processing unit constituting vehicle periphery monitoring device Flow chart showing object classification processing Explanatory drawing about the object area Explanatory drawing regarding the method for determining the presence or absence of closed elements Explanatory drawing about the shape change mode of the object

Explanation of symbols

1. Image processing unit (object area extraction means, object classification means), 2R, 2L, infrared camera (imaging means)

Claims (2)

A vehicle periphery monitoring device that monitors the periphery of a vehicle using a captured image obtained by an imaging means mounted on the vehicle,
An object area extracting means for extracting an area where the object exists from the captured image as an object area;
From the object region extracted by the object region extracting means, a closed element composed of a high luminance pixel group surrounding a low luminance pixel group is different from an open element having a shape in which the lower part of the closed element is cleaved. A vehicle periphery monitoring device comprising: an object classification unit that determines that the object corresponds to the quadruped animal, on the condition that the object is extracted at a time. A vehicle periphery monitoring device that monitors the periphery of a vehicle using a captured image obtained by an imaging means mounted on the vehicle,
An object area extracting means for extracting an area where the object exists from the captured image as an object area;
The requirement is that a closed element composed of a group of high-luminance pixels surrounding a group of low-luminance pixels is intermittently extracted from the object region extracted by the object region extraction means. A vehicle periphery monitoring device comprising: an object classification unit that determines that the animal corresponds to a foot animal.

Images