JP2005339176A - Vehicle recognition device, navigation device and vehicle recognition method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a "vehicle recognition device, a navigation device and a vehicle recognition method" for sharply reducing data quantity to be processed at the time of recognizing the type of a vehicle. <P>SOLUTION: A vehicle recognition processing part 101a calculates a region where the other vehicle is projected based on the position of the other vehicle detected by a radar device 102 to an own vehicle (O) and the width of a road under traveling and the maximum value of height restriction. Also, a vehicle recognition processing part 101a segments the calculated region, and executes image recognition processing to the region for recognizing the type of the projected other vehicle. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a vehicle recognition device, a navigation device, and a vehicle recognition method.

  Conventional navigation devices mounted on vehicles only provide the driver with information such as the traveling position of the host vehicle, the guidance route to the destination, and surrounding landmarks. However, in recent years, more detailed information such as the type of vehicle that travels around the host vehicle, particularly in front or behind, has begun to be demanded.

  Under such circumstances, imaging devices such as CCD (Charge-Coupled Devices) cameras have been reduced in price and have come to be mounted on vehicles as sensors for detecting surrounding vehicles and other road conditions.

  In general, many of the methods for recognizing an object from an image obtained by an imaging device are prepared by extracting features of a correct image given in advance as correct data, and the features of the detection target image and the correct image A so-called pattern recognition method is used in which detection is performed using the correlation with the feature of the pattern. However, the image recognition processing using such feature point extraction processing needs to process an enormous amount of data, and thus the processing cost is often very high.

  Therefore, in the prior art, in order to reduce the processing cost of feature point extraction in vehicle recognition, a method is generally used in which an area in which a vehicle is captured is cut out from an image and only that area is processed. However, this method is often used in conditions where it is easy to limit the processing target region in the originally targeted image, such as vehicle detection using a fixed point camera. On the other hand, for the purpose of recognizing surrounding vehicles by an in-vehicle camera, the camera mounting position is not always constant, and the size of the vehicle that appears in accordance with the distance varies. It is difficult to limit the cut-out area.

As a technique for solving such a problem, for example, in Patent Document 1 shown below, a vehicle is provided with a laser radar and a stereo camera, and the front of the vehicle is scanned by using the laser radar. A technique for detecting a positional relationship with the own vehicle and narrowing down a processing area in an image taken by a stereo camera based on the result is disclosed.
JP-A-7-125567

  However, the prior art disclosed in Patent Document 1 is not for specifying the type of vehicle. For this reason, it is not configured to determine a processing region in which the entire vehicle is included.

  Accordingly, the present invention has been made in view of the above problems, and provides a vehicle recognition device, a navigation device, and a vehicle recognition method capable of significantly reducing the amount of data to be processed when recognizing the type of vehicle. The purpose is to provide.

  In order to achieve such an object, the present invention is an object recognition device that recognizes a type of a vehicle by cutting out a region where the vehicle is projected from an image of the vehicle and recognizing the region. The area on which the vehicle is projected is cut out as a processing target area based on the position of the vehicle, the width of the road, and the maximum value of the vehicle height limit with respect to the position where the imaging means for capturing the image is installed. The By cutting out an area where the vehicle is projected and performing image processing only on this area, the amount of data to be processed when recognizing the type of vehicle can be greatly reduced. Further, by using a constant value in the real world and applying the road width and the vehicle height limit to the constant, the vehicle can be surely placed in the cutout region.

  The vehicle recognition apparatus according to the present invention includes a radar unit for detecting a vehicle, a shooting unit for shooting the vehicle, a position of the vehicle detected by the radar unit with respect to a position where the shooting unit is installed, and a road on which the vehicle is running. Region extraction means for cutting out an area in which the vehicle is projected from the image obtained by the photographing means based on the width of the vehicle and the maximum vehicle height limit, and recognition processing of the area cut out by the area extraction means Vehicle recognition means for recognizing the type of vehicle. By cutting out an area where the vehicle is projected and performing image processing only on this area, the amount of data to be processed when recognizing the type of vehicle can be greatly reduced. Further, by using a constant value in the real world and applying the road width and the vehicle height limit to the constant, the vehicle can be surely placed in the cutout region.

  In addition, the vehicle recognition device includes a condition reading unit that reads a road width of a road on which the vehicle is traveling and a maximum height limit set on the road from a predetermined storage unit, and the region extraction unit includes The area to be cut out may be calculated based on the road width read by the condition reading means and the maximum vehicle height limit. By changing the cutout area according to the conditions set for the road, it is possible to reliably include the vehicle in the cutout area while minimizing the image area to be recognized.

  In addition, the navigation device of the present invention includes a radar unit that detects vehicles around the own vehicle, an imaging unit that captures the surroundings of the own vehicle, a position of the vehicle detected by the radar unit with respect to the own vehicle, and a road that is running Region extraction means for cutting out an area in which the vehicle is projected from the image obtained by the photographing means based on the width of the vehicle and the maximum vehicle height limit, and recognition processing of the area cut out by the area extraction means Vehicle recognition means for recognizing the type of vehicle. By cutting out the area where the vehicle is projected and performing image processing only on this area, the amount of data to be processed when recognizing the type of vehicle traveling around the host vehicle can be greatly reduced. It becomes possible. Further, by using a constant value in the real world and applying the road width and the vehicle height limit to the constant, the vehicle can be surely placed in the cutout region.

  The navigation device may include navigation means for presenting a result recognized by the vehicle recognition means to a driver. By using the recognition result for navigation, the driver can recognize the surrounding situation accurately.

  The vehicle recognition method according to the present invention includes a first step of detecting a vehicle, a second step of shooting the vehicle with a shooting unit, and a position of the vehicle detected in the first step where the shooting unit is installed. A third step of cutting out an area in which the vehicle is projected from the image obtained by the photographing means based on the position, the width of the road being traveled, and the maximum value of the vehicle height limit, and an area cut out in the third step And a fourth step of recognizing the type of vehicle by performing recognition processing. By cutting out an area where the vehicle is projected and performing image processing only on this area, the amount of data to be processed when recognizing the type of vehicle can be greatly reduced. Further, by using a constant value in the real world and applying the road width and the vehicle height limit to the constant, the vehicle can be surely placed in the cutout region.

  ADVANTAGE OF THE INVENTION According to this invention, the vehicle recognition apparatus, navigation apparatus, and vehicle recognition method which can reduce significantly the data amount which should be processed at the time of recognizing the kind of vehicle are realizable.

  Hereinafter, the best mode for carrying out the present invention will be described in detail with reference to the drawings.

  First, Embodiment 1 according to the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram illustrating a configuration of a car navigation apparatus 100 according to the present embodiment. As shown in FIG. 1, a car navigation apparatus 100 includes a CPU (Central Processing Unit) 101, a radar apparatus 102, a CCD (Charge Coupled Devices) camera 103, a DVD (Digital Versatile Disk) drive 111, and a GPS (Global Positioning System). The reception unit 112, the self-contained navigation sensor 113, the transmission / reception unit 114, the operation unit 115, the display unit 116, and the speaker 117 are included. These units are communicably connected via a bus 118.

  In the above configuration, the DVD drive 111 is configured to read a recording medium such as a DVD (Digital Versatile Disk) -ROM storing map data and other guide data. However, in the present invention, other storage means capable of storing map data and other guidance data such as a CD (Compact Disk) -ROM (Read Only Memory) drive and a hard disk drive are applied. be able to.

  The GPS receiving unit 112 is configured to receive a GPS signal transmitted from a GPS satellite and detect the latitude and longitude of the current position of the host vehicle (O). The detected current position information is input to the CPU 101 and used for predetermined processing in navigation.

  The self-contained navigation sensor 113 includes an angle sensor 113a including a gyro for detecting the vehicle direction, and a distance sensor 113b that generates a pulse for every predetermined travel distance. Detect speed. The transmission / reception unit 114 is configured by an in-vehicle telephone for communicating with various service centers.

  The operation unit 115 is configured to allow the user to input various operations / settings to the navigation device 100. This may be configured individually such as a remote controller device or a control panel, or may be configured as a touch panel type input device integrally with a display unit 116 described later. Furthermore, it may be constituted by a microphone for voice input.

  The display unit 116 is configured by, for example, a liquid crystal display type display device or the like, and is configured to display various information such as a map, a guide route, a current position of a vehicle, and other landmarks to the user. The guidance route is a route to the destination that is actually provided to the user, and is a route that is superimposed on the map image on the display unit 116. The speaker 117 is configured to output voice guidance information and the like to the user. In addition, the speaker 117 can also output music input from an audio device or the like.

  In addition, the CPU 101 executes various programs read from the DVD drive 111, and thus, the vehicle existing around the own vehicle (O) based on the information input from the radar device 102 and the CCD camera 103 together with the navigation function. A recognition function (hereinafter referred to as a vehicle recognition function) is realized. A configuration for realizing these functions is referred to as a vehicle recognition device 110 in this description, and will be described in detail below with reference to the drawings.

  FIG. 2 is a block diagram illustrating a configuration of the vehicle recognition device 110. As shown in FIG. 2, the vehicle recognition device 110 includes a radar device 102, a CCD camera 103, and a CPU 101. The CPU 101 implements the vehicle recognition processing unit 101a and the navigation processing unit 101c by reading a predetermined program from the DVD drive 111, for example.

  The radar apparatus 102 may be a radar apparatus that can detect the distance to the vehicle (hereinafter referred to as a partner vehicle) around the host vehicle (O), the direction of the partner vehicle, and the like, such as a millimeter wave radar. Anything can be applied. The radar apparatus 102 is preferably attached around the own vehicle (O), particularly in front and rear, and is attached so as to be able to detect a partner vehicle existing around the own vehicle (O). Therefore, for example, when the radar apparatus 102 has directivity, the antenna of the radar apparatus 102 can be rotated, or at least one radar apparatus 102 is provided at the front and rear.

  It is preferable that the CCD camera 103 is attached so as to be able to photograph a region that is substantially the same as the region that can be detected by the radar device 102. Therefore, the CCD camera 103 is also attached so that the photographing direction can be rotated, or at least one CCD camera 103 is provided at the front and rear. Note that the center line of the region detectable by the radar apparatus 102 and the optical axis of the CCD camera 103 are adjusted so as to substantially coincide. Thereby, it is possible to easily derive the region on the image on which the opponent vehicle is projected from the detection result.

  Detection information acquired by the radar apparatus 102 and an image acquired by the CCD camera 103 (hereinafter referred to as an acquired image) are input to the vehicle recognition processing unit 101 a of the CPU 101 via the bus 118. Note that, as described above, the detection information includes information on the distance of the opponent vehicle from the host vehicle (O) and the direction in which the opponent vehicle exists (hereinafter simply referred to as a direction).

  The vehicle recognition processing unit 101a executes a process of recognizing the opponent vehicle existing in the detectable area from the own vehicle (O) based on the input detection information and the acquired image. This recognition process is repeatedly executed in a predetermined cycle.

  Next, the operation of the vehicle recognition processing unit 101a will be described with reference to the drawings. FIG. 3 is a flowchart showing the operation of the vehicle recognition processing unit 101a. As shown in FIG. 3, the vehicle recognition processing unit 101a inputs detection information and an acquired image (steps S101 and S102). Here, an example of the information included in the detection information is shown in FIG. 4, and an example of the acquired image is shown in FIG. As can be read from FIG. 4 and the above-described content, the detection information includes information such as the distance and direction from the own vehicle (O) to the opponent vehicle. Further, as can be read from FIG. 5, the opponent vehicle (C1) is photographed in the acquired image. For the sake of simplification of description, the following description focuses on only an acquired image obtained by photographing the front of the host vehicle (O) (the lane region in FIG. 4).

  When the detection information and the acquired image are input as described above, the vehicle recognition processing unit 101a calculates the distance L to the opponent vehicle from the detection information (step S103), and also detects the direction of the opponent vehicle, that is, radar detection. The direction of the position is specified (step S104).

  Next, the vehicle recognition processing unit 101a calculates a region (hereinafter referred to as a cutout region) on the acquired image in which the opponent vehicle is projected from the calculated distance L and the specified direction (step S105), and cuts out this region. (Step S106).

  Here, the calculation method of the cutout region will be described in more detail. In calculating the cutout area, in this embodiment, fixed values in the real world are used for the width w and height h of the cutout area. That is, in this embodiment, when the width w of the cut-out area is converted into a scale in the real world, it is calculated from a value that is, for example, about 3 m (meters), which is a general lane width. Further, when the height h of the cut-out area is converted into a scale in the real world, for example, it is calculated from a value that is approximately 4 m, which is a generally limited maximum vehicle height. As described above, by using a constant value in the real world, in the present embodiment, the opponent vehicle can be surely placed in the cutout region. In the following description, the width in the real world is W and the height is H.

Next, an expression for calculating the width w and height h of the cutout region is shown. First, a procedure for calculating the width w will be described. Here, as shown in FIG. 6, if the optical axis of the CCD camera 103 is the x axis and the detected direction of the opponent vehicle is inclined by φ _{1 with} respect to the optical axis of the CCD camera 103, the own vehicle (O ) To the opponent vehicle along the optical axis is expressed by the following formula 1.

Further, when the spread of the imaging area of the CCD camera 103 from the optical axis is θ ₁ , the horizontal length E _W of the real world projected in front of the distance D from the own vehicle (O), that is, the acquired image is displayed. The total length E _W of the landscape is expressed by the following formula 2.

Therefore, the ratio R _w that the constant width W occupies in the horizontal width E _W of the acquired image is expressed by Equation 3 below.

Here, if the number of horizontal bits of the entire acquired image is B _W , the horizontal number of bits b _w of the cut-out area is expressed by Equation 4 below. In case of rounding, round down or round off.

Next, a procedure for calculating the height h of the cutout area will be described. As shown in FIG. 7, when the spread from the optical axis of the imaging area of the CCD camera 103 is θ ₂ , the vertical length E _H of the real world projected in front of the distance D from the own vehicle (O), that is, acquired. The overall length E _H of the landscape projected in the image is expressed by the following formula 5.

Therefore, the ratio R _H of the height H of the area is a constant occupied in height E _H of the acquired image becomes Equation 6 below.

Here, if the vertical bit number of the entire acquired image is B _H , the vertical bit number b _h of the cutout region is expressed by the following Expression 7. In case of rounding, round down or round off.

As described above, when the horizontal bit number b _w and the vertical bit number b _h of the cut-out area are obtained, the vehicle recognition processing unit 101a then determines the size of this size with reference to the radar detection position as shown in FIG. Assign an area to the acquired image and cut out the enclosed area.

In this embodiment, it is assumed that the optical axis of the CCD camera 103 is oriented in the horizontal direction. Therefore, the height H ₂ in the real world from the ground to the radar detection position is a fixed value. On the other hand, the height H ₁ (= H−H ₂ ) from the radar detection position height H ₂ to the maximum vehicle height H is also a fixed value. Therefore, the coordinates (x, y) based on the upper left pixel of the radar detection position in the cutout region are expressed by the following Expression 8.

  Returning to FIG. When the area where the opponent vehicle is projected is cut out as described above, the vehicle recognition processing unit 101a then obtains the correlation between the feature of the correct image given in advance and the feature in the cut-out area for the cut-out area. As a result, the type of vehicle displayed (sedan, light vehicle, wagon, minivan, 1BOX, etc.) is specified (step S107). That is, the vehicle type is specified by performing pattern recognition. Thereafter, the vehicle recognition processing unit 101a outputs the recognition result to the navigation processing unit 101c (step S108).

  When the recognition result is input as described above, the navigation processing unit 101c notifies the driver of the input result according to necessity or purpose. Further, by performing the same configuration and operation on the rear side of the own vehicle (O), it is possible to recognize the opponent vehicle existing behind the own vehicle (O).

  With the configuration and operation as described above, according to the present embodiment, it is possible to greatly reduce the amount of data to be processed when recognizing the type of vehicle traveling around the vehicle, particularly forward or backward. Therefore, the recognition result can be obtained quickly.

  In the present invention, the real world width W and height H used when calculating the width w and height h of the cutout region in the above configuration may be set for each road. This value is accumulated in the map data used in the navigation device 100, and information such as road width and vehicle height restrictions is added for each road. Therefore, it is possible to realize processing accurately corresponding to road conditions (also referred to as road conditions) by reading out these data according to the travel route of the host vehicle (O) and using it for calculation. It becomes.

Further, in the above configuration, a constant in the real world (width W) is adopted as the width w of the cut-out area. However, the present invention is not limited to this, and for example, an interval between lanes (white lines, etc.) in the acquired image is displayed as an image. It may be obtained by processing, and this may be used as the width b _w of the cutout region. In this case, the calculation process for calculating the width b _w is omitted.

  In addition, the said Example 1 is only an example for implementing this invention, This invention is not limited to these, It is in the scope of the present invention to modify these Examples variously, and It is obvious from the above description that various other embodiments are possible within the scope of the present invention.

It is a block diagram which shows the structure of the car navigation apparatus 100 by Example 1 of this invention. It is a block diagram which shows the structure of the vehicle recognition apparatus 110 by Example 1 of this invention. It is a flowchart which shows operation | movement of the vehicle recognition process part 101a. 3 is a diagram illustrating an example of an outline of information included in detection information acquired by a radar apparatus 102. FIG. It is a figure which shows an example of the acquired image acquired with CCD camera 103. FIG. It is a figure for demonstrating the concept at the time of calculating the width w of the cut-out area | region in Example 1 of this invention. It is a figure for demonstrating the concept at the time of calculating the height h of the cut-out area | region in Example 1 of this invention. It is a figure which shows the specific example of the cut-out area | region calculated in Example 1 of this invention.

Explanation of symbols

100 navigation device 101 CPU
101a Vehicle recognition processing unit 101c Navigation processing unit 102 Radar device 103 CCD camera 111 DVD drive 112 GPS receiving unit 113 Autonomous navigation sensor 113a Angle sensor 113b Distance sensor 114 Transmission / reception unit 115 Operation unit 116 Display unit 117 Speaker 118 Bus C1 Counter vehicle O Self vehicle

Claims (6)

An object recognition device that recognizes the type of the vehicle by cutting out an area in which the vehicle is projected from an image of the vehicle and recognizing the area,
A region in which the vehicle is projected is cut out as a processing target region based on the position of the vehicle, the width of the road, and the maximum value of the vehicle height limit with respect to the position where the photographing unit that photographed the image is installed. Object recognition device. Radar means for detecting the vehicle;
Photographing means for photographing the vehicle;
Based on the position of the vehicle detected by the radar unit with respect to the position where the imaging unit is installed, the width of the road on which the vehicle is running, and the maximum height limit, the imaging unit obtains an area where the vehicle is projected. Area cutting out means for cutting out from the captured image;
Vehicle recognition apparatus comprising vehicle recognition means for recognizing a type of vehicle by performing recognition processing on the area cut out by the area cutout means. Condition reading means for reading out the width of the road on which the vehicle is traveling and the maximum value of the vehicle height limit set on the road from a predetermined storage means,
3. The vehicle recognition apparatus according to claim 2, wherein the area extraction unit calculates the area to be extracted based on a road width read by the condition reading unit and a maximum vehicle height limit. Radar means for detecting vehicles around the own vehicle;
Photographing means for photographing the surroundings of the own vehicle;
Based on the position of the vehicle detected by the radar means relative to the own vehicle, the width of the road on which the vehicle is running, and the maximum value of the vehicle height limit, the region where the vehicle is displayed is cut out from the image obtained by the imaging means. Area extraction means;
A navigation apparatus comprising vehicle recognition means for recognizing the type of vehicle by performing recognition processing on the area cut out by the area cutout means.   5. The navigation apparatus according to claim 4, further comprising navigation means for presenting a result recognized by the vehicle recognition means to a driver. A first step of detecting a vehicle;
A second step of photographing the vehicle with the photographing means;
Based on the position of the vehicle detected in the first step with respect to the position where the imaging means is installed, the width of the running road, and the maximum value of the vehicle height limit, an area where the vehicle is projected is displayed on the imaging means. A third step of cutting out from the obtained image;
A vehicle recognition method comprising: a fourth step of recognizing the type of vehicle by performing a recognition process on the region cut out in the third step.

Images