JP2007233440A - On-vehicle image processor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To set an image area for carrying out image processing for correctly associating an image acquired in a predetermined scan timing with the position of an object detected by a radar. <P>SOLUTION: This on-vehicle image processor is provided with a radar 3 for detecting an object ahead of its own vehicle by beam scan; a camera 1 to be driven synchronously with this for picking up the image of the front part of its own vehicle; and a speed sensor for detecting the speed of its own vehicle. Information acquired from those radar 3, camera 1 and speed sensor is input to an image processing area setting part 5, and an image area for performing image processing for every detected object. When the detected object does not exist at the corresponding position in the image, an image processing area setting part 5 resets an image area in a virtual space formed by connecting the position of the object with an endless apoapsis set in the image based on a moving distance where its own vehicle has moved. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an in-vehicle image processing apparatus that is provided in an in-vehicle device that performs speed control, stop control, and the like of a host vehicle and that processes an image ahead of the host vehicle acquired by an imaging device.

  A radar that scans the front of the host vehicle with an electromagnetic beam (laser beam, etc.) and detects an object such as a preceding vehicle is put into practical use as an in-vehicle device for monitoring the front of the host vehicle and contributing to inter-vehicle distance control. In addition, an imaging device that improves the recognition accuracy of a preceding vehicle and other objects by acquiring an image including an object detected by a radar by capturing an image of the front of the host vehicle and performing image processing has been put into practical use. ing.

  The imaging device is driven in synchronization with the scanning of the radar, compares the position of the object acquired by the radar and the captured image, and associates the position with the image to set the image area. Then, image processing is performed on this image area. In this way, setting the image area prevents a reduction in processing speed caused by processing all images, and enables high-speed image processing for each beam scanning period.

A conventional image processing apparatus having such a configuration is disclosed in Patent Document 1, for example.
Japanese Patent No. 3264060

  In the apparatus having the above configuration, in order to associate the position of the object acquired by the radar with the image, the scene when measured by the radar and the scene when captured by the imaging apparatus need to be at the same timing. This is because if the timing of the two is greatly shifted, the scene measured by the radar and the scene measured by the imaging device are different, and the position of the object is different in both scenes. This will be described with reference to FIGS.

  FIG. 1 shows a first scene during the first scan of the radar. FIG. 2A shows an image of the first scene, and FIG. 2B shows a beam scan position in the first scene by radar coordinates. FIG. 2 shows a second scene during the last scan of the radar. FIG. 6A shows an image acquired during the first scan of the radar, and FIG. 6B shows the beam scan position in the second scene in radar coordinates.

  As shown in FIGS. 1B and 2B, the beam is scanned in a fan shape from FIG. 1B to FIG. 2B, and one scan takes about 50 msec. Further, the image pickup apparatus is driven at the first scan timing in FIG. 1, and an acquired image at that time is stored in a memory, and the image is used for image processing.

  1 and 2, a region P is a field of view set in advance, and image processing is performed within this range. The beam BM1 output from the radar indicates the position and shape of the beam emitted at the first scan timing. A beam BM2 indicates the position and shape of the beam emitted at the last scan timing. In FIG. 2A, image areas Q1 to Q3 are areas set for performing image processing, and the black circle reflection points indicating the positions where the beams are reflected are the same if they have substantially the same distance and direction. The areas are grouped as if they were reflected from the object, and each area is set by associating the grouped reflection points in the image. In the figure, an image area Q1 indicates an area surrounding a fixed object beside the lane detected by the beam, and an image area Q2 indicates an area surrounding a preceding vehicle traveling on the lane. An image area Q3 indicates an area surrounding the pedestrian.

  In FIG. 1B, the grouped gray reflection point indicates the position of the object obtained by the previous beam scan. In FIG. 2, the grouped black circle reflection point indicates the current beam scan. The position of the object obtained in is shown.

  As is clear from the above figures, in the first scene shown in FIG. 1, the first scan timing and the imaging timing of the imaging device are the same, but in the second scene shown in FIG. Since 50 msec has elapsed, there may occur a case where the correspondence between the position of the reflection point detected by the reflection of the beam at the last scan timing and the image cannot be obtained. That is, as shown in FIG. 2, the image acquired by the imaging device is 50 msec before, even though the position of the pedestrian is moving toward the own vehicle while the own vehicle is moving for 50 msec. For this reason, the position of the pedestrian captured in the first scene is shifted from the position of the pedestrian detected by the radar in the second scene after 50 msec (actual position). . On the other hand, for the preceding vehicle, the position of the reflection point detected by the radar does not change because the same inter-vehicle distance as that of the host vehicle is maintained after 50 msec. Further, since the fixed object is detected at the first scan timing, the position is not different from the position detected at the last scan timing of the previous time. Note that FIG. 2B shows that only the pedestrian moves the reflection point detected in the previous scan to the near side (own vehicle side) in the current scan.

  For the above reasons, for objects such as pedestrians that hardly move with respect to the scan time, the correspondence between the actual position acquired by the radar and the position on the image at the timing when a fixed time has elapsed from the initial scan timing There was a problem that the relationship was incorrect.

  Further, when the imaging apparatus is driven at the last scan timing, the correspondence between the position of the reflection point of the object detected at the first scan timing and the image cannot be obtained. FIG. 3 shows this state, and shows that a correspondence relationship with respect to a fixed object existing on the side of the road cannot be taken. The same problem occurs even when the imaging apparatus is driven at an intermediate scan timing.

  An object of the present invention is to provide an in-vehicle image processing apparatus that corrects an image area for performing image processing so that an image acquired at a predetermined scan timing and an object position detected by a radar correctly correspond to each other. is there.

The present invention provides a radar that scans the front of the host vehicle with an electromagnetic wave beam to detect an object such as a preceding vehicle, and captures an image including the object by imaging the front of the host vehicle at a predetermined scanning timing for each scan of the radar. An imaging device to acquire, and a host vehicle travel information acquisition unit that acquires host vehicle travel information such as the vehicle speed of the host vehicle,
An in-vehicle image processing device that obtains an image area to be processed in the image based on an object detected by the radar and an image acquired by the imaging device at the predetermined scanning timing is provided.

  An example of the electromagnetic wave beam is a laser beam, and the radar scans the beam from left to right so as to form a fan shape in front of the vehicle. The beam can be scanned two-dimensionally. In this case, the scanning position in the vertical direction is moved downward while repeating the scanning from left to right.

  The imaging device is composed of, for example, a CCD camera with a global shutter, and captures an image of the front at the moment when the shutter is opened, and stores this in a memory. The shutter is opened at a predetermined scan timing of the beam, and this operation is performed for each scan. The predetermined scan timing is, for example, the first scan timing.

  The vehicle travel information acquisition unit measures the speed (vehicle speed) of the host vehicle, the yaw rate, and the like.

  The in-vehicle image processing apparatus according to the present invention obtains an image area to be processed in the image based on an object such as a preceding vehicle detected by the radar and the image. In the image processing, the feature amount of an object in the image area is extracted and the object is recognized.

  With the above configuration, the present invention includes image area resetting means for resetting the image area.

  The image area resetting means, when an object detected by the radar until the end of one scan does not exist at the corresponding position in the image, the host vehicle obtained by the host vehicle travel information and the predetermined scan timing The image area is reset in a virtual space formed by connecting the position of the object detected by the radar and the infinity point set in the image based on the movement distance moved by.

  By such an operation, the image area is expanded, and an object can be detected in the area. When this resetting is performed, it is necessary to take into account movements such as the speed of the host vehicle. This is because if the vehicle is at high speed, the area of the image area that is reset in the virtual space must be larger, and if the vehicle rotates along the lane, the image along the curvature This is because it is necessary to set an area.

  The image area resetting means sets a first image area having a constant size at the position of the object detected by the radar, and connects the corner of the first image area and the point at infinity. Form a virtual space.

The infinity point is referred to as an image vanishing point (hereinafter simply referred to as vanishing point) in the present invention. Generally, this vanishing point is set as a fixed position at the center position of the screen. The first image area is set to a rectangular shape obtained by an arbitrary method based on the position of the object detected by the radar, or a rectangular shape having a predetermined size based on the same position. By connecting the corners and vanishing points of the first image area, a quadrangular prism virtual space is formed. In this virtual space, a second image area is set at the position of the travel distance traveled by the host vehicle obtained from the host vehicle travel information and the predetermined scan timing, and the first image area and the second image area are set. The area obtained by combining the image areas along the edge of the virtual space is reset as the image area. As a result, the reset image area has a larger area on the image coordinates than the image area before resetting. Moreover, since the area is formed based on information such as the speed of the host vehicle, the probability that an image of an object exists in the area is very high.

  According to the present invention, when the position of the object detected by the radar is shifted from the position of the object in the image captured before that, the object is included in the image area for image processing. Since the image area is reset so as to include the image, the image processing on the object detected by the radar is always performed correctly. For this reason, there is an advantage that the accuracy of image processing is improved.

  FIG. 4 is a schematic block diagram of an in-vehicle image processing apparatus that is an embodiment of the present invention.

  The camera 1 is composed of a CCD camera or the like with a global shutter, and is provided in the vicinity of the rear mirror of the host vehicle with the imaging surface facing forward. This camera 1 images the front of the host vehicle.

  The entire image storage unit 2 stores an image captured by the camera 1.

  The radar 3 is configured by a laser radar, emits a laser beam (hereinafter simply referred to as a beam) in front of the host vehicle, and receives the reflected beam. In this embodiment, one scan is completed by reciprocating the left and right beams of an ellipse with a narrow width on the irradiation surface, thereby searching for a front two-dimensional range.

  The signal processing unit 4 measures the distance to the object by measuring the round trip time of the beam light based on the signal from the radar 3. By measuring the distance at each scan timing, all the objects in front can be detected within one scan, and the distance to these objects can be measured. Further, signals from a vehicle speed sensor and a yaw rate sensor (not shown) are received and sent to the image processing area setting unit 5 described later. A steering angle sensor can be used instead of the yaw rate sensor.

  The trigger signal generator 6 receives the scan start signal from the radar 3 and generates a trigger signal. This trigger signal is output to the imaging command unit 7, where a shutter-on signal for the camera 1 is generated and output to the camera 1. Therefore, the camera 1 acquires a forward image when the shutter is turned on at the timing when the radar 3 starts the first scan.

  The image stored in the entire image storage unit 2 is input to the image processing area setting unit 5. Further, the vehicle speed from the vehicle speed sensor and the yaw rate from the yaw rate sensor are input to the image processing area setting unit 5 via the signal processing unit 4.

  As will be described later, the image processing area setting unit 5 sets and resets an image area. Partial images in each image area set for one or more entire images are extracted, stored in the partial image storage units 1 to N, and image processing is performed in the image processing units 1 to N. In the image processing unit, the shape and moving direction of the image are recognized, and each of the objects 1 to N has higher-order processing devices (for example, inter-vehicle distance control device) as position information with attribute information (attribute information such as shape and moving direction). Is output.

  FIG. 5 is a configuration diagram of the signal processing unit 4.

  The signal processing unit 4 calculates the distance and direction to the object based on the signal from the radar 3, and sends the distance and direction data to the image processing area setting unit 5 via the interface 41. Signals from the vehicle speed sensor and the yaw rate sensor are input to the vehicle travel information acquisition unit 42, and the curvature radius of the road is calculated based on the travel speed obtained by the vehicle speed sensor and the data obtained from the yaw rate. The signals and calculation results from each sensor acquired here are sent to the image processing area setting unit 5 via the interface 41.

  FIG. 6 is a configuration diagram of the image processing area setting unit 5.

  The image processing area setting unit 5 includes a white line recognition unit 50 that recognizes a white line on a road, a vanishing point calculation unit 51, and an area setting unit 52. As described above, the vanishing point means an infinite point in the image. The white line recognition part 50 is used when calculating | requiring a vanishing point in Example 2 mentioned later. The area setting unit 52 sets and resets an image area. When the vanishing point is set as a fixed point in advance, the area setting unit 52 sets and resets the image area without using the output of the vanishing point calculation unit 51. When the vanishing point is obtained based on the white line recognized by the white line recognition unit 50, the area setting unit 52 sets and resets the image area based on the vanishing point.

  Next, the operation will be described.

  7 and 8 are flowcharts showing the schematic operation of the in-vehicle image processing apparatus.

  FIG. 7 shows the operation of the radar 3.

  In step (ST) 1, a radar scan is started. The state of FIG. 1 corresponds to this timing. Next, ranging to an object by scanning is performed in ST2. Subsequently, it is determined whether or not the timing is the timing of the trigger position of the camera 1. In the present embodiment, since the first scan timing is set to the trigger position timing, the process proceeds from ST1 to ST2 to ST3 as the first scan timing. If it is determined in ST3 that it is the first scan timing corresponding to the trigger position, a trigger signal is generated in the trigger signal generator 6 in ST4, and the shutter of the camera 1 is turned on. In ST5, the process proceeds to the next scan, and thereafter, ranging is continuously performed in ST2 until all scans are completed. Since the trigger signal is generated only at the first scan timing, the camera 1 acquires an image only once at the first scan timing for each scan.

  In ST 6, the vehicle speed and yaw rate are acquired, and this data is output to the image processing area setting unit 5. In ST7, ranging to an object (object) is performed for each scan based on data obtained in all scans. In ST8, the distance measurement data is output to the image processing area setting unit 5.

  FIG. 8 shows operations of the camera 1 and the image processing area setting unit 5.

  In ST10, when the camera 1 receives a shutter-on signal from the imaging command unit 7, the shutter is turned on, the process proceeds to ST11, imaging is started, and the captured image is acquired and stored in the entire image storage unit 2 (main memory) ( ST12). Subsequently, when ranging data is received from the signal processing unit 4 (ST13), the image area is calculated in ST14.

  The image area is set as follows.

  As shown in FIG. 1, after acquiring an image at the first scan timing and receiving ranging data for one scan from the radar, is the object detected for each scan present at the corresponding position in the image? If it exists, an image area is set around the area where the beam reflection points are grouped. In FIG. 2, the image areas Q1 and Q2 are set around the area where the beam reflection points are grouped because the fixed object detected by the radar 3 and the preceding vehicle are present at the corresponding positions in the image. It is an image area.

  On the other hand, when the object detected by the radar 3 does not exist at the corresponding position in the image, the image area is set by the method shown in FIG. FIG. 7 is a diagram for explaining the first embodiment for setting an image area. In the first embodiment, the road is a straight road, and an infinite point in the image is set as a fixed vanishing point S. Now consider the pedestrian at the right end of the road.

  As shown in the figure, a first image area Q30 is set with reference to an area obtained by grouping points detected by the radar. The first image area Q30 has a predetermined height and width (or can be automatically set based on the grouped area) and is set in a rectangular shape. The corner of the first image area Q30 and the vanishing point S, which is an infinite point of the image, are connected to form a virtual space X having a quadrangular prism shape. The vanishing point S, which is an infinite point, is set as a fixed point as previously described. Then, from Q30 toward the vanishing point S which is a fixed point, the vehicle speed of the host vehicle is moved by a distance corresponding to a value obtained by multiplying the vehicle speed by 50 msec (time from the first scan timing to the last scan timing), A rectangular second image area Q31 is formed in the virtual space at the moved position. Next, the first image area and the second image area are combined. The combined image area is reset as a new image area. FIG. 4B shows the reset image area Q32. As shown in the figure, the reset image area Q32 is larger than the image area Q30 before resetting. As a result, the image area reset from Q30 to Q32 includes an image captured by the camera 1 at the first timing (in this example, a pedestrian).

  In the flowchart of FIG. 8, when an image area is set in ST14 (including an image area to be reset), images included in the image area are transferred to partial image storage units 1 to N (sub-memory) in ST15. The By repeating ST14 to ST16 for all the objects (objects) detected by the radar (ST16), the images in the respective image areas are transferred to the partial image storage units 1 to N, respectively.

  In ST17 to ST19, predetermined image processing is performed on the images in the partial image storage units 1 to N.

  In the calculation for setting the image area, it is necessary to match the coordinates of the radar 3 and the camera 1 and project them to the image coordinates. Specifically, the calculation is performed as follows.

  As shown in FIG. 10, the coordinate systems of the camera 1 and the radar 3 are (XC, YC, ZC) and (XL, YL, ZL), respectively. Then, the relationship between the two coordinates is as shown in Equation 1.

  Expression 1 is an expression for converting laser coordinates into camera coordinates. Expression 2 is an expression for projecting camera coordinates to image coordinates.

  Therefore, the radar coordinates can be projected to the image coordinates through the camera coordinates by Expression 1 and Expression 2. These equations are common in computer vision technology. For example, “Three-dimensional vision” (2004, Kyoritsu Shuppan, written by Xu Tsuyoshi and Saburo Saburo).

  FIG. 11 shows the radar coordinates two-dimensionally. Now, it is assumed that the position of the pedestrian (object) detected by the radar 3 has moved relatively as shown (actually due to the movement of the host vehicle). A rectangle is set based on the center of gravity of the area where the detection points after movement detected at the last scan timing are grouped, the width w, and the height H (predetermined fixed value). (In FIG. 11, only B1 and B2 are shown because they are shown in two dimensions, but actually B3 and B4 exist at positions perpendicular to the paper surface. The same applies to other rectangles. .) B1 to B4 on the radar coordinates are coordinate-converted from Equations 1 and 2 to obtain b1 to b4. On the image coordinates, b1 to b4 are plotted as shown in FIG. The b1-b4 and the vanishing point S are connected to form a quadrangular prism virtual space.

  Next, the travel distance vT is obtained by multiplying the scan time T from the trigger generation position (first scan timing) to the rectangles B1 to B4 (last scan timing) and the vehicle speed v of the host vehicle. A rectangle having apexes A1 to A4 is set at a position moved in the moving direction of the host vehicle by vT. The size of the rectangle is the same as the rectangle with vertices B1 to B4. Then, A1 to A4 on the radar coordinates are coordinate-converted from Equations 1 and 2 to obtain a1 to a4. When a1 to a4 are plotted on the image coordinates, the result is as shown in FIG. As a result, a rectangle having corners a1 to a4 is set in a virtual space of a quadrangular prism connecting b1 to b4 and the vanishing point. FIG. 12 shows the rectangles b1 to b4 obtained by coordinate conversion of the rectangles B1 to B4 using Equations 1 and 2, and the rectangles a1 to a4 obtained by performing coordinate conversion of the rectangles A1 to A4 using Equations 1 and 2. Yes. Both indicate positions on image coordinates. FIG. 13 shows an image area that is reset by connecting the image areas set as rectangles a1 to a4 and the image areas set as rectangles b1 to b4 in FIG. a1 to a4 and b1 to b4 are coupled along the ridgeline of the virtual space.

  The specific calculation for resetting the image area is performed as described above.

  FIG. 14 is an explanatory diagram of the second embodiment for setting an image area.

  In the second embodiment, the road is straight, but the vanishing point S is not set as a fixed point, but the vanishing point S is dynamically set for each frame (every scan) according to the shape of the lane. . That is, a white line before the lane (a white line or a center line formed on the side of the lane) is recognized from the image acquired by the camera 1, and the intersection of the white lines is set as a vanishing point. The white line in the foreground is used so that the vanishing point can be easily obtained even when the lane is curved, when there is a preceding vehicle, when the road is far away or downhill. is there. The white line can be recognized easily from the image acquired by the camera 1 and can be performed by a general image recognition process. As one specific recognition method, for example, in the “lane keep assist system” "Development of white line recognition system" (Vol.12, No.1 HONDA R & D Technical Review 2000, written by Hirota, Kajiura, and Ikeda). The operation procedure other than setting the vanishing point S is exactly the same as in the first embodiment.

FIG. 15 is an explanatory diagram of the third embodiment for setting an image area.
In the third embodiment, the road is curved and the vanishing point S is set as a fixed point.

  Most curved parts of expressways and automobile roads are designed as clothoid curves. The clothoid curve is a curve in which the product of the curvature radius and the curve length is constant, and is used as a relaxation curve for smoothly connecting roads having two different curvature radii. The coordinate values drawn by the clothoid curve can be expressed as Equation 3 and Equation 4 using the radius of curvature R and the curve L (however, an approximate expression with the higher order terms omitted).

  Expressions 3 and 4 are obtained when the object (object) moves from the position A5 to A8 (only A5 and A6 are shown in the figure) to the position B5 to B8 (only B5 and B6 are shown in the figure). This represents a change in coordinate value that changes to.

  The radius of curvature R is obtained by substituting the angular velocity ω measured from the detection signal of the yaw rate sensor into Equation 5. The curve length L from the position of A5, A6 in FIG. 16 to the position of B5, B6 is obtained from Equation 6 assuming that the traveling vehicle speed v is constant during this time.

Therefore, Equation 3 and Equation 4 are obtained from Equation 5 and Equation 6. By obtaining Equations 3 and 4, the position of the object moves from B5 and B6 to A5 and A6 as in the first embodiment. A5, A6 and B5, and B6 are transformed into a5 to a8 and b5 to b8 by coordinate transformation according to equations 1 and 2, resulting in an image area as shown in FIG. 17, and the image area is reset as shown in FIG. The
FIG. 19 is an explanatory diagram of the fourth embodiment for setting an image area.

  In the fourth embodiment, the road is curved, and the vanishing point S is dynamically set according to the road shape. In the fourth embodiment, no yaw rate sensor is used.

  A plurality of vanishing points are used. Here, the case where two vanishing points S1 and S2 are used will be described.

  First, the vanishing point S1 is obtained at the straight line portion of the white line in front of the host vehicle. The vanishing point S2 is obtained by the tangent of the curve portion further forward in the traveling direction of the host vehicle. As in the first embodiment, a first virtual space is formed between B9 to B12 and the vanishing point S1 on the radar coordinates in FIG. 19, and the moving distance vT is set in the moving direction of the host vehicle in this space. A rectangular image area having apexes A9 to A12 is set at the moved position. Similarly, a second virtual space is formed between B9 to B12 and the vanishing point S2, and in this space, the vertices are C9 to C12 whose vertices are moved in the moving direction of the host vehicle by the moving distance vT. Set the image area. Of the rectangular image area with vertices A9 to A12 and the rectangular image area with vertices C9 to C12, an image area located far from the vehicle on the radar coordinates is selected, and this image area The image area is reset by combining the image areas B9 to B12. FIG. 20 shows the image area before resetting, and FIG. 21 shows the reset image area.

  With the above control operation, for straight roads and curved roads, the vanishing point is set to a fixed value (Examples 1 and 3), or the vanishing point is obtained by white line recognition (Examples 2 and 4). Can be reset.

  In the above embodiment, the predetermined scan timing is set as the first scan timing, and the trigger timing of the camera 1 is matched with the first scan timing. However, the trigger timing of the camera 1 may be matched with the last scan timing. It may be good or may be coincident with the scanning timing in the middle. When the trigger timing of the camera 1 is made coincident with the last scan timing, the image area Q31 in FIG. 9 is set on the own vehicle side with respect to Q30. That is, a virtual space is formed by extending a line connecting the vanishing point S and the corner of the image area Q30 to the own vehicle side, and the image area Q31 is set in this space.

  Furthermore, although the image area is rectangular, it may be any shape such as a polygonal shape or an elliptical shape.

A first scene during the first scan of the radar is shown. A second scene at the last scan of the radar is shown. A scene in the case of imaging at the last scan of the radar is shown. 1 is a block diagram of an in-vehicle image processing apparatus that is an embodiment of the present invention. It is a block diagram of a signal processing part. It is a block diagram of an image processing area setting part. It is a flowchart which shows operation | movement of the radar of the vehicle-mounted image processing apparatus which is embodiment of this invention. It is a flowchart which shows operation | movement of the camera and image processing area setting part of the vehicle-mounted image processing apparatus which is embodiment of this invention. It is explanatory drawing of Example 1 which performs the setting of an image area. It is a figure which shows the attachment position of a camera and a radar, and each coordinate system. It is a figure which shows the coordinate of the radar of Example 1 in two dimensions. 6 is a diagram illustrating a specific image area setting method according to the first exemplary embodiment. FIG. FIG. 10 is a diagram for explaining resetting of an image area according to the first embodiment. It is explanatory drawing of Example 2 which performs the setting of an image area. It is explanatory drawing of Example 3 which performs the setting of an image area. It is a figure which shows the coordinate of the radar of Example 3 in two dimensions. FIG. 10 is a diagram illustrating a specific image area setting method according to the third embodiment. FIG. 10 is a diagram illustrating resetting of an image area according to a third embodiment. It is a figure which shows the coordinate of the radar of Example 4 which performs the setting of an image area in two dimensions. FIG. 10 is a diagram illustrating a specific image area setting method according to a fourth embodiment. FIG. 10 is a diagram for describing resetting of an image area according to a third embodiment.

Claims (6)

A radar that scans the front of the host vehicle with an electromagnetic wave beam to detect an object such as a preceding vehicle, and an imaging device that captures an image of the front of the host vehicle at a predetermined scanning timing and acquires an image including the object for each scan of the radar And a host vehicle travel information acquisition unit that acquires host vehicle travel information such as the vehicle speed of the host vehicle,
In an in-vehicle image processing apparatus for obtaining an image area to be processed in the image based on an object detected by the radar and an image acquired by the imaging device at the predetermined scanning timing,
If an object detected by the radar until one scan is completed does not exist at the corresponding position in the image, the travel distance traveled by the host vehicle obtained by the host vehicle travel information and the predetermined scan timing is obtained. And an image area resetting means for resetting the image area in a virtual space formed by connecting the position of the object detected by the radar and the infinity point set in the image. An in-vehicle image processing apparatus.   The image area resetting means sets a first image area having a constant size at the position of the object detected by the radar, and connects the corner of the first image area and the point at infinity. The virtual space is formed, a second image area is set in the virtual space at the position of the moving distance, and the first image area and the second image area are combined along the edge of the virtual space. The in-vehicle image processing apparatus according to claim 1, wherein the set area is reset as the image area.   The in-vehicle image processing apparatus according to claim 2, wherein the image area resetting unit sets the infinity point as a fixed point in advance.   The in-vehicle image processing apparatus according to claim 2, wherein the image area resetting unit recognizes a white line of a lane in the image and sets an extension point of the white line as the infinity point. The host vehicle travel information acquisition unit acquires the vehicle speed and angular velocity of the host vehicle and the radius of curvature of the lane as host vehicle travel information,
The in-vehicle image processing apparatus according to claim 3, wherein the image area resetting unit sets the second image area at a position moved by the moving distance of the host vehicle along the radius of curvature based on the information.   The image area resetting means recognizes a white line of a lane in the image, sets an extension point of a straight line portion of the white line as a first infinity point, and sets an extension point of a tangent line of the curved line portion of the white line as a first point. 2 as an infinite point, a second image area candidate is set for each infinite point, and an image area candidate farthest from the first image area is set as a second image area Item 2. The in-vehicle image processing apparatus according to Item 2.

Images