JP5166933B2 - Vehicle recognition device and vehicle - Google Patents

Description

The present invention relates to an apparatus for recognizing an object using an image acquired via an imaging unit, and a vehicle .

  2. Description of the Related Art Conventionally, a technique is known in which a periphery is imaged by an imaging unit and an object existing in the periphery is recognized from the captured image (for example, see Patent Document 1). With this technology, for example, an object existing on and around a road on which the host vehicle is traveling can be recognized, and information presentation to the driver and vehicle traveling control can be performed according to the recognition result.

The stop lamp lighting detection device disclosed in Patent Document 1 detects a stop lamp of a preceding vehicle by applying a technique for detecting a lighting lamp using color information and luminance information of a pixel from a color image obtained by imaging the front of the host vehicle. To do. Specifically, in this apparatus, a region in which a specific color (stop lamp color) and its approximate color are continuous for a certain number of pixels or more is extracted from the image, and the degree of approximation from the extracted region to the specific color of each pixel is determined. A region where the average value is equal to or greater than the threshold is specified, and if the average luminance value of the specified region is equal to or greater than the threshold, it is determined as a stop lamp. Thereby, the preceding vehicle is recognized. The stop lamp lighting detection device is combined with an inter-vehicle distance approach warning device, and an approaching state with the preceding vehicle is evaluated based on the stop lamp detection result of the preceding vehicle, and an alarm is issued.
JP-A-8-335300

  As described above, when a vehicle is detected by using a stop lamp (brake lamp) of the vehicle, it is necessary to distinguish the brake lamp of each vehicle when there are a plurality of vehicles in the vicinity. However, in the apparatus of Patent Document 1, no consideration is given to the case where there are a plurality of vehicles. For this reason, in the apparatus of Patent Document 1, when a plurality of vehicles are captured in an overlapping manner and there are a plurality of brake lamp groups in the image, the brake lamps of each vehicle cannot be distinguished and detected. There is an inconvenience that it is not recognized.

The present invention has been made in view of such a background, and even when there are a plurality of vehicles in the vicinity, a vehicle recognition device that can distinguish between a plurality of vehicles captured in an image and appropriately recognize the vehicle. And to provide a vehicle .

In order to achieve such an object, the vehicle recognition device of the present invention is a vehicle recognition device that recognizes other vehicles existing in the vicinity based on an image acquired through an imaging unit mounted on the host vehicle. A feature region extraction unit configured to extract, as a feature region, a region corresponding to an image portion of a brake lamp of a preceding vehicle, the pixel having a pixel value within a predetermined range from an image acquired through the imaging unit; Among the feature areas extracted by the area extraction means, the pairing means for specifying the first feature area and the second feature area arranged in the horizontal direction on the image, and the pairing means specified by the pairing means for the first feature area in the surrounding area of the in the feature region of the first second characteristic region direction, the first determination region corresponding to the image portion of the body portion around the brake lamp of a leading vehicle A determination region setting for setting the second determination region at a position symmetrical to the first determination region with respect to the center line of the first and second feature regions. A feature that determines that the first and second feature regions are image portions of one other vehicle when the degree of correlation between the means and the first determination region and the second determination region is high. An area determining means is provided (first invention).

According to the vehicle recognition system of the first invention, the feature region extraction unit, from the image captured a peripheral, Ri Do from the pixel having a pixel value of a predetermined range, you corresponds to an image portion of the brake lamp of a leading vehicle region Are extracted as feature regions. Thereby, a region corresponding to the image portion of the brake lamp is extracted as a feature region. At this time, when there are a plurality of vehicles in the vicinity, feature regions for the plurality of vehicles are mixed on the image.

In contrast, brake lamps, since the two provided symmetrically with respect to the center in the vehicle width direction of the vehicle (horizontal direction), and thus arranged in the horizontal direction on the image. Therefore, the pairing means identifies a set of two feature regions (first feature region and second feature region) arranged in the horizontal direction on the image. As a result, a set that is highly likely to correspond to a pair of brake lamps of one vehicle is specified. However, the group specified in this way may include a case where two vehicles are arranged in the horizontal direction and a brake lamp of one vehicle and a brake lamp of the other vehicle are combined.

Therefore, the determination region setting means uses the first feature region corresponding to the image portion of the vehicle body portion around the brake lamp of the preceding vehicle in the peripheral region in the second feature region direction in the first feature region . The second determination region is set at a position symmetrical to the first determination region with respect to the center line of the first and second feature regions. When a set of two characteristic areas corresponds to a pair of brake lamps of one vehicle, these determination areas correspond to a body part around the brake lamp of the same vehicle. In this case, since the body parts of the same vehicle have many parts having the same brightness and color, the features of the first and second determination areas on the image are similar, and the first determination area and the second determination area are similar to each other. The degree of correlation with the area increases.

On the other hand, when the set of two feature areas is composed of brake lamps of different vehicles, if the determination area is set in the peripheral area in the center line direction of the two feature areas, the determination area is a background portion other than the vehicle. It corresponds to. In this case, since various things are imaged in the background portion, the characteristics of the first and second determination areas on the image are different, and the degree of correlation between the first determination area and the second determination area is different. Lower . Therefore, by feature region determining means, when the degree of correlation between the first determination area and the second determination area is high, the first and second characteristic region in the image portion of one other vehicle By determining that there is a plurality of vehicles, even if there are a plurality of vehicles in the vicinity, the plurality of vehicles captured in the image can be distinguished and the vehicles can be appropriately recognized.

  In the vehicle recognition device of the first invention, it is preferable that the feature region extraction means extracts a region composed of pixels having a luminance value equal to or greater than a predetermined value as the feature region (second invention).

  According to the vehicle recognition device of the second invention, since the brake lamp or the like is generally high in luminance, the brake lamp or the like can be obtained by extracting a region composed of pixels having a luminance value equal to or higher than a predetermined value by the feature region extraction means. Can be extracted.

  Alternatively, in the vehicle recognition device according to the first aspect of the present invention, when the image acquired through the imaging unit is a color image, the feature region extraction unit is a range in which a color component corresponds to a specific color as the feature region. It is preferable to extract a region composed of the pixels to be (third invention).

  According to the vehicle recognition device of the third invention, since the brake lamp or the like is generally a specific color such as red or yellow, an area composed of pixels whose color components fall within a range corresponding to the specific color by the characteristic area extraction means. The feature region corresponding to the brake lamp or the like can be extracted.

  Further, in the vehicle recognition device according to any one of the first to third inventions, the pairing means is arranged in the horizontal direction on the image, and the shape of one feature region and the other feature among the feature regions. It is preferable to specify the first feature region and the second feature region in which the degree of similarity with the shape obtained by mirror-reversing the region in the horizontal direction is a predetermined value or more (fourth invention).

  According to the vehicle recognition device of the fourth aspect of the invention, for example, the pair of brake lamps of one vehicle is provided symmetrically with respect to the center in the vehicle width direction (horizontal direction) of the vehicle. In a set of two feature regions corresponding to a pair of brake lamps of a vehicle, it is considered that the degree of similarity between the shape of one feature region and the shape obtained by mirror-reversing the other feature region in the horizontal direction is high. Therefore, the first feature in which the degree of similarity between the shape of one feature region and the shape of the other feature region mirror-reversed in the horizontal direction is equal to or greater than a predetermined value by the pairing means in the horizontal direction on the image. By specifying the region and the second feature region, a set of two feature regions corresponding to a pair of brake lamps or the like of one vehicle can be specified with higher accuracy among the feature regions.

  Moreover, in the vehicle recognition device according to any one of the first to fourth inventions, the feature area determination unit is configured to calculate an average of pixel values of the first determination area and an average of pixel values of the second determination area. When the difference is less than or equal to a predetermined value, it is preferable to determine that the degree of correlation between the first and second determination areas is high (fifth invention).

  According to the vehicle recognition device of the fifth aspect of the invention, the average pixel value of the first and second determination areas indicates the tendency of the characteristics of each determination area, and therefore the difference between the averages of these pixel values is small. In this case, it is understood that the degree of correlation between the first and second determination regions is high. As the pixel value, for example, the luminance value of the pixel or the color component of the pixel can be used when the image is a color image. Therefore, when the average difference between the pixel values of the first and second determination areas is equal to or less than the predetermined value by the characteristic area determination unit, it is determined that the degree of correlation is high, whereby a set of two characteristic areas is obtained. It can be appropriately determined that the vehicle corresponds to one other vehicle.

Next, the vehicle of the present invention is a vehicle that includes an imaging unit and has a function of recognizing other vehicles existing in the vicinity based on an image acquired through the imaging unit.
A feature region extraction unit configured to extract, as a feature region, a region corresponding to an image portion of a brake lamp of a preceding vehicle, the pixel having a pixel value within a predetermined range from an image acquired through the imaging unit; Among the feature areas extracted by the area extraction means, the pairing means for specifying the first feature area and the second feature area arranged in the horizontal direction on the image, and the pairing means specified by the pairing means For the first feature region, a first determination region corresponding to the image portion of the vehicle body portion around the brake lamp of the preceding vehicle is set in the peripheral region in the direction of the second feature region in the first feature region . And determining region setting means for setting the second determination region at a position symmetrical to the first determination region with respect to the center line of the first and second feature regions. And the above Characteristic area determining means for determining that the first and second characteristic areas are image parts of one other vehicle when the degree of correlation between the determination area and the second determination area is high. (6th invention).

According to the vehicle of the sixth invention, as described for the vehicle recognition device of the first invention, the pairing means identifies the first feature region and the second feature region arranged in the horizontal direction on the image. By doing so, a set of two characteristic regions that are highly likely to correspond to a pair of brake lamps of one vehicle is specified. Then, by the determination area setting means, the first and second determination areas are centered in the peripheral area in the center line direction of the two characteristic areas in the first and second characteristic areas with respect to the set of the two characteristic areas . By setting the position symmetrical with respect to the line, the degree of correlation between the first determination area and the second determination area is high, and thus a set of two characteristic areas is a pair of one vehicle. It can be determined that it corresponds to a brake lamp. Therefore, when the degree of correlation between the first determination area and the second determination area is high by the characteristic area determination means, the first and second characteristic areas are image portions of one other vehicle. By determining, even if there are a plurality of vehicles in the vicinity, the plurality of vehicles captured in the image can be distinguished and other vehicles existing in the vicinity can be appropriately recognized.

  An embodiment of the present invention will be described with reference to FIGS. Referring to FIG. 1, vehicle recognition device 10 is used by being mounted on host vehicle 1 (the vehicle of the present invention), and the surroundings of host vehicle 1 from an image captured by camera 2 that images the front of host vehicle 1. Is detected, and the relative position between the other vehicle and the host vehicle 1 is recognized. The camera 2 (imaging means of the present invention) is a CCD camera or the like, and is attached to the front portion of the host vehicle 1.

The data of the relative position of the other vehicle and the host vehicle 1 that is recognized by the vehicle both recognition device 10 is output to the vehicle 1 of the ECU (Electronic Control Unit) 20. The ECU 20 determines the possibility that the host vehicle 1 will come into contact with the other vehicle based on the relative position between the other vehicle and the vehicle 1, and the situation where the host vehicle 1 may be in contact with the other vehicle. The warning output process for performing a warning output (by voice output from a speaker (not shown), etc.) and the braking of the own vehicle 1 so that the contact between the own vehicle 1 and another vehicle is avoided. A contact avoidance control process for assisting driving of the steering is executed.

  Next, referring to FIG. 2, the vehicle recognition device 10 functions as an image acquisition unit 11, a feature region extraction unit 12, a pairing unit 13, a determination region setting unit 14, and a feature region determination unit 15. And.

  The image acquisition unit 11 receives an image signal output from the camera 2 and acquires a color image composed of pixel data from the input image signal. At this time, the pixel data is composed of color components composed of an R value, a G value, and a B value.

  The feature area extraction unit 12 extracts, from the color image acquired by the image acquisition unit 11, an area including pixels whose pixel values fall within a predetermined range as a feature area. At this time, the feature region extraction unit 12 uses a color component of the pixel as a pixel value, and extracts a region composed of pixels in which the color component is in a range corresponding to a specific color (red in the present embodiment). Specifically, the feature region extraction unit 12 determines that the color component falls within a range corresponding to the specific color using the hue / saturation calculated from the color component (R, G, B) of the pixel. .

  The pairing unit 13 includes a feature region extracted by the feature region extraction unit 12 and arranged in a horizontal direction on the image, and a shape of one feature region and a shape obtained by mirror-inverting the other feature region in the horizontal direction. A first feature region and a second feature region in which the degree of similarity is greater than or equal to a predetermined value are specified.

  The determination region setting unit 14 sets the first determination region in the peripheral region in the second feature region direction for the first feature region specified by the pairing unit 13. At the same time, the determination area setting unit 14 has the second feature area specified by the pairing unit 13 at a position symmetrical to the first determination area with respect to the center line of the first and second feature areas. A second determination area is set.

  When the degree of correlation between the first determination region and the second determination region set by the determination region setting unit 14 is high, the feature region determination unit 15 determines whether the first and second feature regions are different from each other. It determines with it being the image part of a vehicle. Specifically, the feature area determination unit 15 determines whether the first and first characteristics are different when the difference between the average pixel value of the first determination area and the average pixel value of the second determination area is equal to or less than a predetermined value. It is determined that the degree of correlation between the two determination areas is high. At this time, the feature area determination unit 15 uses the luminance value calculated from the color components (R, G, B) of the pixel as the pixel value.

Car both recognition device 10 and the A / D converter circuit for converting an input analog signal into a digital signal, an image memory for storing the image signal obtained by digitizing, accessing data stored in the image memory (read and write) a interns face circuit for aggregation, the image memory stored in a computer for performing various arithmetic processing on the image (CPU, memory, arithmetic operation circuit consisting of input and output circuits, etc., or these functions An electronic unit composed of a microcomputer. Vehicle recognition device 10, by executing the car both recognition program in the computer, the computer is, the feature region extraction unit 12, the pairing means 13, the determination region setting unit 14, as a characteristic region judging means 15 Function.

  Next, the operation (vehicle recognition processing) by the vehicle recognition device 10 will be described according to the flowchart shown in FIG. The process according to the flowchart shown in FIG. 3 is repeatedly executed every control cycle of the vehicle recognition apparatus 10.

In STEP 1 of FIG. 3, the image acquisition unit 11 acquires a color image via the camera 2. The acquired color image is A / D converted and stored in the image memory. FIG. 4A illustrates a color image I ₁ obtained by the camera 2 at a certain control cycle time. The example shown in FIG. 4A is an example in the case where a preceding vehicle T exists in front of the host vehicle 1.

  Next, in STEP 2, the feature area extraction unit 12 uses the following formulas (1) and (2) from the color components (R, G, B) of the pixels of the color image acquired in STEP 1 to display the pixels in the HSV color space. The hue H and saturation S are calculated.

  Note that MAX is the maximum value of (R, G, B), and MIN is the minimum value of (R, G, B).

Next, in STEP 3, the feature region extraction unit 12 uses the calculated hue H and saturation S to extract a feature region including pixels whose color component is in a range corresponding to red. Specifically, the feature region extraction unit 12 _adds a hue H within a predetermined range H _{th1 to} H _th2 and a saturation S within a predetermined range S _{th1 to} S _th2 to a color image acquired via the camera 2. The pixel to be “1” (white) and the other pixels “0” (black) are subjected to a binarization process, and a labeling process or the like is further performed to extract a feature region. The predetermined ranges H _{th1 to} H _th2 and S _{th1 to} S _th2 are determined in advance as ranges corresponding to red.

FIG. 4B illustrates an image I ₂ obtained by extracting a feature region from the image I ₁ illustrated in FIG. In FIG. 4B, hatched areas are extracted feature areas. In the example shown in FIG. 2B, characteristic areas A1 and A2 corresponding to the brake lamp of the preceding vehicle T and a characteristic area A3 corresponding to the stop lamp of the preceding vehicle T are extracted.

Next, in STEP 4, the pairing means 13 is similar to the shape obtained by aligning the extracted feature regions in the horizontal direction on the image, and the shape of one feature region and the other feature region mirror-inverted in the horizontal direction. A set of two feature regions (first feature region and second feature region) in which the degree A is equal to or greater than a predetermined value _Ath is specified. The predetermined value _Ath is determined in advance as a range of similar degrees that a set of two feature areas corresponding to a pair of brake lamps of one vehicle can take on an image. Moreover, when attaching the camera 2 to the own vehicle 1 so that the front of the own vehicle 1 may be imaged like this embodiment and recognizing the other vehicle which exists in front of the own vehicle 1 (front and diagonally forward), The X and Y directions of the image correspond to the horizontal and vertical directions in real space (world coordinate system), respectively.

In the example shown in FIG. 4 (b), the pairing means 13, as the feature region are arranged in the horizontal direction, for any horizontal line L _H in the image are included in the range R of a predetermined width W including a horizontal line L _H Two feature areas A1 and A2 are specified. Further, the pairing means 13 calculates the degree of similarity A between the shape of the feature region A1 and the shape (A2 ′ in FIG. 4B) obtained by mirror-inverting the feature region A2 in the horizontal direction. Since the degree A is equal to or greater than the predetermined value _Ath , the first and second feature regions A1 and A2 are specified as a set of two feature regions.

  Thereby, n sets of two feature regions are specified. For example, in the example shown in FIG. 4B, one set of two feature regions (first feature region A1 and second feature region A2) is specified.

Next, the processing of STEP 5 to 10 is executed for each identified group. In the following, the first feature region of each set is represented by A1 _i , and the second feature region is represented by A2 _i (i = 1,..., N).

First, in STEP 5, the determination area setting unit 14 sets the first determination area B1 _i in the peripheral area (position separated by a predetermined distance) in the second feature area A2 _i direction of the first feature area A1 _i . At the same time, the second determination region B2 _i is set in the peripheral region in the direction of the first feature region A1 _{i of} the second feature region A2 _i (position separated by the same predetermined distance). The first determination area B1 _i and the second determination area B2 _i are in a line-symmetric position with respect to the center line L _{v of} the first and second feature areas A1 _i and A2 _i . The image I ₃ in FIG. 4C shows the first and second determination regions B1 and B2 set for the first and second feature regions A1 and A2 shown in FIG. 4B. As shown in FIG. 4C, when the first and second characteristic areas A1 and A2 correspond to a pair of brake lamps of the preceding vehicle T, the first and second determination areas B1 and B2 are: This corresponds to the body part around the brake lamp of the preceding vehicle T.

Next, in STEP 6, the feature area determination unit 15 calculates the luminance value Y of the pixel from the color components (R, G, B) of the pixel of the color image acquired in STEP 1, and stores the luminance value Y of the first determination area B 1 _i . A difference ΔY between the average of the luminance values and the average of the luminance values of the second determination region B2 _i is calculated. Specifically, the feature area determination unit 15 calculates the luminance value Y of the pixel by Y = α × R + β × G + γ × B from the color components (R, G, B) of the pixel of the acquired color image. However, α, β, and γ are predetermined coefficients such that α + β + γ = 1. Note that the feature region determination unit 15 may calculate the luminance value Y by using Y = (MAX + MIN) / 2 by using the maximum value MAX and the minimum value MIN among the R value, G value, and B value. Alternatively, the G value may be used as the luminance value Y.

Next, in STEP 7, the feature area determination unit 15 determines whether or not the average difference ΔY of the calculated luminance values is equal to or less than a predetermined value ΔY _th . When the determination result in STEP 7 is YES (the average difference ΔY in luminance values is equal to or less than the predetermined value ΔY _th ), the correlation between the first determination area B1 _i and the second determination area B2 _i is high, and the first , Second determination areas B1 _i , B2 _i can be determined to correspond to the vehicle body portion of the same vehicle. Therefore, it can be considered that the first and second feature areas A1 _i and A2 _i correspond to a pair of brake lamps of the same vehicle. In this case, proceeding to STEP 8, the feature region determination means 15 calculates the center coordinates P _i of the first and second feature regions A1 _i and A2 _i in the image coordinate system. The center coordinates P _i corresponds to the barycentric coordinates of the other vehicle. In the example shown in FIG. 4C, center coordinates P ₁ corresponding to the barycentric coordinates of the preceding vehicle T are calculated from the first and second feature areas A1 and A2.

Next, in STEP 9, the feature area determination means 15 performs coordinate conversion of the center coordinates P _i into the world coordinate system. Specifically, the camera parameters such as the focal length and pixel interval of the camera 2, the size of the first and second feature areas A1 _i and A2 _i on the image, and the size that the brake lamp can take in the real space The coordinates are converted based on the distance of the other vehicle to the host vehicle 1 estimated from the above. Thereby, in STEP10, the relative position with respect to the own vehicle 1 of another vehicle is acquired by the vehicle recognition apparatus 10, and the other vehicle is recognized.

On the other hand, when the determination result in STEP 7 is NO (the average difference ΔY in luminance values is not equal to or less than the predetermined value ΔY _th ), the correlation between the first determination region B1 _i and the second determination region B2 _i is low, It is considered that the first and second feature areas A1 _i and A2 _i do not correspond to a pair of brake lamps of the same vehicle. In this case, the process returns to STEP 5 and the processing of STEPs 5 to 10 is performed for the next group.

  When a plurality of other vehicles are present by executing the above-described processing of STEP 5 to 10 for each group, a set of two feature regions corresponding to a pair of brake lamps of each other vehicle is specified, The relative position of each other vehicle with respect to the host vehicle 1 can be acquired.

  The above is the vehicle recognition process in the vehicle recognition apparatus 10 of this embodiment. According to this embodiment, even when there are a plurality of vehicles in the vicinity, the plurality of vehicles captured in the image can be distinguished and the vehicle can be appropriately recognized.

  In the present embodiment, the feature region extraction unit 12 uses red as the specific color. However, as another embodiment, a specific color other than red (for example, yellow) may be used. Further, the feature region extraction unit 12 extracts the feature region using the hue / saturation, but values other than the hue / saturation may be used.

  In the present embodiment, the feature region extraction unit 12 extracts a region (specific color region) composed of pixels in which the color component is in a range corresponding to a specific color as the feature region. As an example, a region (high luminance region) composed of pixels having a luminance value equal to or higher than a predetermined value may be extracted.

  Further, in the present embodiment, the pairing means 13 is similar to the shape obtained by aligning the extracted feature regions in the horizontal direction on the image, and the shape of one feature region and the other feature region mirror-inverted in the horizontal direction. The combination of two feature areas whose degree is greater than or equal to a predetermined value is specified. However, as another embodiment, a set of two feature areas arranged in the horizontal direction on the image is specified without using a shape. It may be a thing.

  In the present embodiment, the determination area setting unit 14 sets the first and second determination areas in the peripheral area in the center line direction of the first and second feature areas. The first and second determination areas may be set in the upper and lower peripheral areas of the first and second feature areas.

  In the present embodiment, the feature region determination unit 15 determines the degree of correlation between the first and second determination regions when the average difference between the luminance values of the first and second determination regions is equal to or less than a predetermined value. However, it is also possible to determine the degree of correlation by another method. The feature area determination unit 15 may determine the degree of correlation between the first and second determination areas using a value other than the luminance value, such as hue and saturation.

  In the present embodiment, a color image is acquired via the camera 2, but a grayscale image in which pixel data is composed of luminance values may be acquired. In this case, the feature region extraction unit 12 extracts the feature region using the luminance value, and the feature region determination unit 15 determines the degree of correlation between the first and second determination regions using the luminance value.

  In the present embodiment, the image acquisition unit 2 is configured such that the color component of the pixel data is composed of the R value, the G value, and the B value. However, as the color component of the pixel data, for example, a CMY output or the like is used. It is good.

  In the present embodiment, the front of the host vehicle 1 is imaged by the camera 2 and the vehicle recognition device 10 recognizes the other vehicle in front of the host vehicle 1. The vehicle recognition device 10 may recognize other vehicles behind the host vehicle 1 using a pair of headlamps of the vehicle as a characteristic part of the vehicle.

  As another embodiment, the host vehicle 1 includes a radar (for example, a millimeter wave radar, an ultrasonic radar, a laser radar, or the like) that detects an object, and the position information of the object obtained as a detection result of the radar is obtained from an image. It is good also as what acquires the relative position of an other vehicle and the own vehicle by matching with the recognized other vehicle.

Explanatory drawing which showed the mounting aspect to the own vehicle of the vehicle recognition apparatus of this invention. The block diagram of the vehicle recognition apparatus shown in FIG. The flowchart which shows the vehicle recognition process which recognizes the other vehicle by the vehicle recognition apparatus shown in FIG. Explanatory drawing of the vehicle recognition process of FIG.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Vehicle (own vehicle), 2 ... Camera (imaging means), 10 ... Vehicle recognition apparatus, 11 ... Image acquisition means, 12 ... Feature area extraction means, 13 ... Pairing means, 14 ... Determination area setting means, 15 ... Feature area determination means, 20... ECU.

Claims (6)

In the vehicle recognition device for recognizing other vehicles existing in the vicinity based on the image acquired through the imaging means mounted on the host vehicle,
From the image acquired through the imaging means, a feature area extraction means for extracting, as a feature area, an area corresponding to an image portion of a brake lamp of a preceding vehicle, which consists of pixels having a pixel value within a predetermined range;
Pairing means for specifying a first feature area and a second feature area arranged in a horizontal direction on the image among the feature areas extracted by the feature area extraction means;
The first feature region specified by the pairing means corresponds to the image portion of the vehicle body portion around the brake lamp of the preceding vehicle in the peripheral region in the direction of the second feature region in the first feature region . A first determination area to be set and a second determination of the second feature area at a position symmetrical to the first determination area with respect to the center line of the first and second feature areas Determination area setting means for setting an area;
Feature region determination means for determining that the first and second feature regions are image portions of one other vehicle when the degree of correlation between the first determination region and the second determination region is high. A vehicle recognition device comprising: The vehicle recognition device according to claim 1,
The vehicle feature recognition apparatus, wherein the feature region extraction unit extracts a region composed of pixels having a luminance value equal to or greater than a predetermined value as the feature region. The vehicle recognition device according to claim 1,
The image acquired through the imaging means is a color image,
The vehicle feature recognition apparatus, wherein the feature region extraction unit extracts a region composed of pixels whose color components correspond to a specific color as the feature region. In the vehicle recognition device according to any one of claims 1 to 3,
The pairing means has a predetermined degree of similarity between the shape of one of the feature regions arranged in the horizontal direction on the image and a shape obtained by mirror-inverting the other feature region in the horizontal direction. A vehicle recognition apparatus that identifies a first feature region and a second feature region. In the vehicle recognition device according to any one of claims 1 to 4,
The feature region determination means determines the first and second determinations when a difference between an average pixel value of the first determination region and an average pixel value of the second determination region is equal to or less than a predetermined value. It is determined that the degree of correlation between regions is high. A vehicle including an imaging unit and having a function of recognizing other vehicles existing in the vicinity based on an image acquired through the imaging unit;
From the image acquired through the imaging means, a feature area extraction means for extracting, as a feature area, an area corresponding to an image portion of a brake lamp of a preceding vehicle, which consists of pixels having a pixel value within a predetermined range;
Pairing means for specifying a first feature area and a second feature area arranged in a horizontal direction on the image among the feature areas extracted by the feature area extraction means;
The first feature region specified by the pairing means corresponds to the image portion of the vehicle body portion around the brake lamp of the preceding vehicle in the peripheral region in the direction of the second feature region in the first feature region . A first determination area to be set and a second determination of the second feature area at a position symmetrical to the first determination area with respect to the center line of the first and second feature areas Determination area setting means for setting an area;
Feature region determination means for determining that the first and second feature regions are image portions of one other vehicle when the degree of correlation between the first determination region and the second determination region is high. A vehicle characterized by comprising: