JP2005309797A - Warning device for pedestrian - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To surely detect a pedestrian in a blind spot to a driver by adequately picking up the image of the blind spot and issue a warning. <P>SOLUTION: A warning device for a pedestrian detects the present position and traveling state of a vehicle, changes an image pickup range around the vehicle in accordance with the present position and the traveling state, processes chronological picked-up images, and detects a pedestrian approaching the vehicle, and then issues a warning. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an apparatus that issues an alarm when there is a possibility that the vehicle is in contact with a pedestrian.

  Even if the direction of the driver's blind spot changes according to the turning angle, the camera automatically points in the blind spot direction to capture the blind spot direction, and the blind spot direction image is displayed on the monitor in the passenger compartment to enter the blind spot range. 2. Description of the Related Art A vehicle surrounding monitoring device is known in which an existing pedestrian or bicycle can be confirmed on a monitor screen (see, for example, Patent Document 1).

Prior art documents related to the invention of this application include the following.
JP 05-238314 A

  However, since the above-described conventional vehicle surrounding monitoring apparatus is configured to only display an image of a blind spot range with poor visibility on the monitor, it is impossible to confirm a pedestrian or bicycle existing in the blind spot range if the monitor image is forgotten. There is a problem.

  The current position and the running state of the vehicle are detected, the imaging range around the vehicle is changed according to the current position and the running state, and the time series images taken are processed to detect and alert to a pedestrian approaching the vehicle.

  According to the present invention, even if the driver's blind spot range changes according to the current position and driving state of the vehicle, the imaging range can be accurately set as the driver's blind spot range, and a pedestrian existing in the blind spot range. Can be reliably detected and alarmed.

  An embodiment will be described in which a pedestrian is detected by processing a captured image of a camera when passing through an intersection with poor visibility, and a warning is issued when there is a possibility of contact.

  FIG. 1 shows the configuration of an embodiment. The camera 201 is attached to the lower part of the left and right front ends of the vehicle, and the direction (the direction of the optical axis of the photographing lens) can be changed by the motor 204. FIG. 1 shows only the camera 201 and the motor 204 installed at the lower right front end of the vehicle, and the illustration of the camera and motor installed at the lower left front end of the vehicle is omitted. The image processing device 202 processes an image captured by the camera 201 to detect a pedestrian, a bicycle, or a vehicle. The navigation device 206 detects the current position of the vehicle by satellite navigation or self-contained navigation. The motor control device 205 drives and controls the motor 204 based on the current position information from the navigation device 206, and changes the direction of the camera 201 to change the imaging range.

  The vehicle speed sensor 207 detects the vehicle speed, and the turn signal switch 208 outputs a right turn signal or a left turn signal according to the operation direction of a turn signal lever (not shown). The brake switch 209 detects the depression state of the brake pedal or the operation state of the hand brake. The parked vehicle confirmation switch 211 is a switch operated when confirming whether a pedestrian or a bicycle exists behind a vehicle parked on the path ahead of the vehicle. The driver operates the parked vehicle confirmation switch 211 when he / she feels anxiety that a pedestrian or bicycle jumps out from behind a vehicle parked in front of the course, and walks that exist behind the parked vehicle Person and bicycle can be confirmed.

  The control device 203 includes a CPU (not shown), a memory, and the like. The control device 203 executes a control program (to be described later) and inputs the vehicle running state from the vehicle speed sensor 207, the turn signal switch 208, and the brake switch 209, and a parked vehicle confirmation switch. A confirmation request for a pedestrian that is behind the parked vehicle of the driver is input from 211, the imaging range of the camera 201 is changed by the motor control device 205, and based on the processing result of the camera image from the image processing device 202 Thus, a pedestrian or bicycle approaching the course of the vehicle is detected, and if there is a possibility of contact with the pedestrian or bicycle, the warning device 210 warns the driver.

  2 to 3 are flowcharts showing a pedestrian warning program according to an embodiment. The operation | movement which determines the danger which contacts with the pedestrian approaching on the course of the own vehicle is demonstrated by these flowcharts. The control device 203 repeatedly executes this pedestrian warning program while an ignition switch (not shown) of the vehicle is turned on.

  In step 301, it is confirmed whether the approach flag to the intersection is ON. In the initial state immediately after the execution of this program is started, the approach flag to the intersection is in an OFF state, and the process proceeds to step 314. In step 314, the navigation device 206 confirms whether or not the vehicle is approaching an intersection. If it is determined that the vehicle is not approaching the intersection, the process proceeds to step 401 where the motor control device 205 is controlled so that the direction of the camera 201 is set to the front of the vehicle by the motor 204.

  FIG. 4 shows the mounting position of the camera, and FIG. 5 shows the imaging range of the camera. The camera 201 is attached to the lower part of the right front end of the vehicle, and can change the direction by the motor 204 to capture an area A in the front of the vehicle, an area B on the side of the vehicle, and an area C on the rear side of the vehicle. Note that the front, side, and rear sides of the vehicle can be similarly imaged with respect to the camera attached to the lower left front end of the vehicle. The orientation of the camera 201 in the initial state is the front of the vehicle. In order to facilitate detection of pedestrians and other vehicles at night, a near infrared lamp may be installed in the vicinity of the camera as shown in FIG.

  Returning to FIG. 3, the explanation of the pedestrian warning operation will be continued. In step 401, the camera 201 is directed to the front of the vehicle, and in step 402, the forward parking stop vehicle recognition switch 211 is turned on, that is, the driver walks behind the vehicle parked on the path ahead of the vehicle. To check if a person or bicycle is present. Details of the case where there is a confirmation request for the presence or absence of a pedestrian behind a parked vehicle in front of the vehicle will be described in detail later. If there is no confirmation request for the presence or absence of a pedestrian behind the parked vehicle in front of the vehicle, the process returns to step 301 in FIG.

  If it is determined in step 314 in FIG. 2 that the vehicle is approaching the intersection, the process proceeds to step 315, the approach flag to the intersection is turned ON, and the process returns to step 301. If the approach flag to the intersection is ON in step 301, the process proceeds to step 302, and whether the vehicle has passed the intersection determined to be approaching the intersection based on the current position information from the navigation device 206 and having the approach flag turned ON. Confirm whether or not. If you have already passed the intersection, go to step 313, assuming that there is no situation where you can detect a pedestrian approaching your vehicle and judge the possibility of contact when you pass the intersection. The approach flag is turned off and the process returns to step 301.

  On the other hand, when it is determined that the vehicle is approaching the intersection and the vehicle has not yet passed the intersection where the approach flag is turned on, the process proceeds to step 303 to determine whether or not the vehicle speed is equal to or lower than the predetermined speed. Usually, since the vehicle travels slowly at the intersection, a speed that can be determined as passing through the intersection is set as the predetermined speed. If the vehicle speed is higher than the predetermined speed, the process returns to step 301. If the vehicle speed is lower than the predetermined speed, the process proceeds to step 304. In step 304, it is determined whether or not the vehicle is near the center of the intersection. If it is determined that the vehicle is near the center of the intersection, as shown in FIG. 9, it is assumed that the vehicle is parked in a state of waiting for a right turn. Although FIG. 9 shows a T-shaped intersection, a similar determination is made in the case of a crossroad intersection.

  If it is determined that the vehicle is near the center of the intersection, it is checked in step 305 whether the turn signal switch 208 on the right turn side is ON. If the turn signal switch 208 on the right turn side is not ON, the process returns to step 301 and the above-described processing is repeated. When the right turn side turn signal switch 208 is ON, it is assumed that the vehicle is waiting for a right turn while looking at the gap between the vehicles facing each other at the center of the intersection. A region C shown in FIGS. 5 and 9 is imaged laterally.

  In step 307, a captured image of the camera 201 is processed to detect a passing vehicle facing each other. In subsequent step 308, it is determined whether or not the detection state of the passing vehicle satisfies a predetermined condition. Here, as a predetermined condition, a determination reference value for the passing frequency of the oncoming vehicle that can be determined that the driver is looking at the right turn is set in advance, and if the passing frequency of the oncoming vehicle is equal to or higher than the determination reference value, the passing It is assumed that the vehicle detection state satisfies a predetermined condition. The details of the method for detecting and determining the passing vehicle will be described later.

  If the detection state of the passing vehicle satisfies the predetermined condition, the process proceeds to step 309, where the captured image of the camera 201 is processed to detect a pedestrian or a bicycle approaching the right turn pedestrian crossing from the right rear side of the vehicle. . The details of this pedestrian inspection method will be described later. In step 310, it is confirmed whether there is a pedestrian or bicycle approaching the right turn pedestrian crossing.

  When the vehicle reaches the intersection and the vehicle speed is less than or equal to the predetermined speed, the vehicle turns to the center of the intersection, the turn signal switch 208 on the right turn side is turned on, and the detection state of the opposite passing vehicle satisfies the predetermined condition, the right turn When it is determined that there are no pedestrians or bicycles approaching the pedestrian crossing, they are waiting for a right turn at the intersection, but there is no possibility of contact with pedestrians or bicycles. You can turn right. In such a situation, the process returns to step 301 and the above-described processing is repeated.

  On the other hand, if it is determined that there is a pedestrian or bicycle approaching the right turn pedestrian crossing in the right turn waiting situation described above, the process proceeds to step 311 to determine whether or not the brake switch 209 has been turned from ON to OFF, that is, a stakeholder. Check whether has released the brake pedal and hand brake. When the driver releases the brake pedal and the hand brake, it is determined that there is a will to start a right turn, and in step 312, the warning device 210 may contact a pedestrian or bicycle on the right turn pedestrian crossing. Warning.

  At the time when the brake pedal and hand brake are released, the automatic transmission vehicle is in a state just before the vehicle starts to move at an extremely low speed due to creeping, so the driver's intention to start a right turn is confirmed in the state of the brake switch 209 Thus, the driver can be alerted at an appropriate timing.

  Next, a process for preventing contact with a pedestrian or bicycle at an intersection without a traffic light will be described. If it is determined in step 304 in FIG. 2 that the point where the vehicle speed is equal to or lower than the predetermined speed is not near the center of the intersection, the process proceeds to step 316, and whether the passing intersection is a no-signal intersection based on information from the navigation device 206. Confirm whether or not. If it is not a no-signal intersection, the process returns to step 301 and the above-described processing is repeated. In the case of a no-signal intersection, the process proceeds to step 317, and the vehicle side area B is imaged as shown in FIGS. 5 and 14, with the direction of the camera 201 at the left and right front ends of the vehicle facing the left and right sides of the vehicle, respectively.

  In step 318, a captured image of the camera 201 is processed to detect a pedestrian on an intersecting road. If it is determined in step 310 that there is a pedestrian on the intersecting road and it is determined in step 311 that the brake switch 209 has been turned from ON to OFF, a warning is issued in step 312.

  Next, a method for detecting a pedestrian or bicycle behind a vehicle parked in front of the vehicle will be described. When it is determined in steps 301 to 314 in FIG. 3 that the own vehicle is not near the intersection, the process proceeds to step 401 in FIG. 4, and the direction of the camera 201 at the left and right front ends is directed to the front of the vehicle. When the driver finds a vehicle stopped on the course ahead of the vehicle as shown in FIG. 15, the parked vehicle confirmation switch 211 is operated in order to detect a pedestrian or bicycle existing behind the vehicle. If the parked vehicle confirmation switch 211 is ON in step 402, the process proceeds to step 403, where the captured image of the camera 201 provided at the lower part of the left and right front ends of the vehicle is processed, and the vehicle is seen from behind the parked vehicle in front. Detect pedestrians and bicycles jumping out on the path.

  When the vehicle existing on the forward path is a small vehicle, the approach of a pedestrian or bicycle can be confirmed through the hood or roof of the vehicle. However, when the parked vehicle is a large vehicle such as a truck, the pedestrian or the bicycle cannot be confirmed behind the vehicle body. However, a large vehicle such as a truck has a high minimum ground clearance, so if the viewpoint of looking forward is lowered, a pedestrian or bicycle behind the vehicle can be seen from under the vehicle. Therefore, as shown in FIG. 15, the movement of the pedestrian's feet seen from under the front parked vehicle is detected from the time-series image captured by the camera 201, and the presence of the pedestrian jumping out on the course of the vehicle is detected. To do. Details of this detection processing will be described later. If it is determined in step 404 that there is a pedestrian or bicycle approaching from the shadow of the forward parked vehicle on the course of the own vehicle, a warning is issued in step 405.

  FIG. 7 is a flowchart showing the passing vehicle detection process executed in step 307 of FIG. The passing vehicle detection process will be described with reference to this flowchart. In step 801, an image of region C captured by the camera 201 is input. When turning right at the intersection, most of the driver's attention is directed to the oncoming vehicle. Therefore, as shown in FIG. 9, the direction of the camera 201 is set to the right rear side of the vehicle, and a region C that supplements the driver's viewing angle is defined. Take an image. As shown in FIG. 8, the general driver's left and right viewing angles are set to about 200 degrees in total, which is 100 degrees left and right with respect to the front of the range that can be seen with both eyes. However, the range in which the driver can watch is only a very small area, and the range in which the colors can be confirmed with both eyes is about 70 degrees. Therefore, it is difficult to check pedestrians and bicycles approaching the right-hand pedestrian crossing in a state where the oncoming vehicles are passing each other.

  After inputting the camera image in step 801 in FIG. 7, it is determined in step 802 whether the vehicle detection initial condition flag is ON. Since the flag is OFF in the initial state, the process proceeds to step 803 to determine whether or not a “difference satisfying a predetermined condition” has occurred in the left end region of the camera image.

  Here, “difference satisfying a predetermined condition” means that each time between images when a passing vehicle enters the imaging range of the camera 201 and a time-series camera image as shown in FIG. 10 is captured. This is a density difference greater than a predetermined value generated in a pixel. Further, hereinafter, a “difference” pixel area where a density difference of a predetermined value or more has occurred is referred to as a “difference occurrence area”.

  When a passing vehicle enters the imaging range of the camera 201, difference areas are generated in order from the first time-series image as shown in FIG. The difference generation area due to the passing vehicle is generated from the left end of the screen, is enlarged to the entire image, and then moves to the right end and disappears. The determination in step 803 captures the initial phenomenon of the difference caused by the passing vehicle. If no difference occurs in the left end area of the image, the process returns to step 801. On the other hand, if a difference that satisfies the predetermined condition occurs in the left end area of the image, the process proceeds to step 804, the vehicle detection initial condition flag is turned ON, and the corresponding image is set as n frames in the memory (not shown) of the control device 203. Remember. Thereafter, the process returns to step 801 to input the next image.

  If there is a difference in the left end area of the camera image, the vehicle detection initial condition flag is ON, so the routine proceeds from step 802 to 805, where the vehicle detection image frame counter is counted up. The vehicle detection image frame counter is a counter configured in a software form of the CPU of the control device 203, and counts the number of image frames in which a difference has occurred.

  In step 806, it is determined whether or not there is a difference generation area that satisfies a predetermined condition in an image frame (n + α) counted up from an image frame n in which a difference first occurs in the left end area of the camera image.

  Here, the predetermined condition of the difference generation region is, for example, based on a reference area (number of pixels) for determining whether or not there is an oncoming vehicle in which the area (number of pixels) of the difference generation region passes on the screen. And the difference occurrence area is moved to the right by a distance corresponding to the count-up number α from the image frame n.

  If it is determined in step 806 that there is no difference occurrence area that satisfies the predetermined condition, the process proceeds to step 807, where the difference in the left end area of the image generated in step 803 does not correspond to a passing vehicle, and vehicle detection is performed. The initial condition flag is turned OFF, the vehicle detection image frame counter is cleared, and the process returns to step 801.

  On the other hand, if it is determined in step 806 that there is a difference generation area that satisfies the predetermined condition, the process proceeds to step 808 to determine whether or not a difference that satisfies the predetermined condition has occurred in the right end area of the image. As described above, the “difference satisfying the predetermined condition” means that when a passing vehicle enters the imaging range of the camera 201 and a time-series camera image as shown in FIG. The density difference of a predetermined value or more generated in each pixel. In the final stage when the vehicle passes through the imaging range of the camera 201, a difference occurs only in the right end region of the image.

  While the difference is generated in the entire image, the process returns to step 801 and the above-described processing is repeated. However, if the difference occurs only in the right end area of the image, the process proceeds to step 809 and the count value of the vehicle detection image frame counter is set to a predetermined value. It is determined whether or not it is greater than or equal to the value. This predetermined value is a criterion value for judging whether or not there is a passing oncoming vehicle, and an appropriate value is set in consideration of the influence of the vehicle type of the passing vehicle, particularly the size and length, the passing speed, etc. To do.

  If the count value of the vehicle detection image frame counter is equal to or larger than the predetermined value, the process proceeds to step 810, and it is assumed that a passing vehicle is detected. The process returns to step 801 and the above-described passing vehicle detection process is repeated. On the other hand, if the count value of the image frame counter is less than the predetermined value, the process proceeds to step 807, the vehicle detection initial condition flag is turned OFF, the image frame counter is cleared, and the process returns to step 801. repeat.

  FIG. 11 shows a difference occurrence state in the camera image when different types of vehicles pass each other. In step 307 of FIG. 2 for determining whether or not the vehicle is waiting for a right turn at the intersection, based on the change in the total number of difference occurrence pixels of each image frame in which the passing vehicle is detected by the passing vehicle detection process described above. Can be determined. The total number of difference occurrence pixels per image capture time interval 1/30 second is measured as vehicle A or D in the case of a passenger car, and is measured as vehicle B in the case of a one-box type vehicle. In the case, it is measured like the vehicle C. Since the time interval at which the difference occurs is known from this measured value, the passing frequency and the inter-vehicle distance of the passing vehicle can be detected, and can be used for determining the right turn waiting state in the intersection.

  FIG. 12 shows a subroutine for detecting an approaching person to the right turn pedestrian crossing. This subroutine is executed in steps 309 and 318 in FIG. 2 and step 403 in FIG. In the pedestrian detection process of FIG. 12, a case will be described in which a pedestrian approaching the pedestrian crossing at the right turn is detected when waiting for a right turn in step 309 in FIG. The method for detecting a person is to detect a pedestrian who comes into contact with the head at a no-signal intersection in step 318 of FIG. 3 and jumps out of the parked vehicle in front of the vehicle in step 403 of FIG. Since the same applies to the detection of a pedestrian coming, explanation of the latter case is omitted.

  In step 1301, a captured image is input from the camera 201, and in a subsequent step 1302, it is confirmed whether or not the approaching person detection initial condition flag is ON. Since the flag is OFF in the initial state, the process advances to step 1303 to determine whether or not a difference area that satisfies a predetermined condition has occurred.

  FIG. 13 is a time-series image of a pedestrian approaching the pedestrian crossing at the right turn taken by the camera 201 from the rear of the vehicle. When there is a pedestrian (including a bicycle) approaching the right turn destination pedestrian crossing, the above-described difference areas are generated in order from the first one in the time-series image shown in FIG. In this case, since it is assumed that a pedestrian approaches from the rear of the vehicle, it occurs in the right end region of the image and sequentially moves to the left end of the image. The area (number of pixels) of the difference area generated in the image is compared with a criterion value set in advance to determine the presence or absence of pedestrians (including bicycles), and the area (number of pixels) of the difference area is determined. If it is equal to or greater than the reference value, it is determined that there is a pedestrian approaching the right-hand pedestrian crossing.

  If no difference satisfying the predetermined condition has occurred in step 1303, the process returns to step 1301, and the camera image is input again. If a difference that satisfies the predetermined condition has occurred and it is confirmed that there is a pedestrian approaching the right-hand pedestrian crossing, the process proceeds to step 1304, and the approaching detection initial condition flag is turned ON. In the subsequent step 1305, the first captured image of the pedestrian approaching the right turn pedestrian crossing is stored as n frames in the memory of the control device 203, and the position where the difference occurs is stored in the memory. Thereafter, the process returns to step 1301 to input the next image.

  If a pedestrian difference area is generated in the camera image, the approaching person detection initial condition flag is ON, so that the process proceeds from step 1302 to 1306 and the approaching person detection image frame counter is counted up. This approaching person detection image frame counter is a counter configured in a software form of the CPU of the control device 203, and counts the number of image frames in which a pedestrian difference area is generated.

  In step 1307, it is determined whether or not there is a difference generation area that satisfies a predetermined condition in an image frame (n + α) counted up from an image frame n in which a difference first occurs in the camera image.

  The predetermined condition is that the area (number of pixels) of the difference generation area is compared with a determination reference value set in advance to determine the presence or absence of a pedestrian (including a bicycle), and the area (pixels) of the difference generation area Number) is equal to or greater than the determination reference value, and the position of the difference generation area on the image frame (n + α) is predetermined from the position of the difference generation area on the image frame n where the difference first occurs in the camera image. The condition is that the image moves leftward within the range. In other words, the conditions are based on the movement characteristics of pedestrians and bicycles.

  If it is determined in step 1307 that there is no difference generation area that satisfies the predetermined condition, the image frame n in which the difference first occurs in the camera image is not an image capturing a pedestrian or a bicycle, that is, a detection target object Since it is not a dangerous object that touches on a pedestrian crossing, the process proceeds to step 1310, the approaching person detection initial condition flag is turned OFF, the approaching person detection image frame counter is cleared, and the process returns to step 1301.

  On the other hand, if it is determined in step 1307 that there is a difference occurrence area that satisfies the predetermined condition, the process proceeds to step 1308 to determine whether or not the count value of the approaching person detection image frame counter is equal to or larger than the predetermined value. If the count value of the image frame counter is equal to or larger than the predetermined value, the process proceeds to step 1309, and the process returns to step 1301 because a pedestrian or bicycle is detected. On the other hand, if the count value of the image frame counter is less than the predetermined value, the process returns to step 1301.

  In Steps 1303 and 1307, when it is determined that there is a range of reference values for determining the presence or absence of pedestrians (including bicycles) and that there are more than a predetermined amount of pedestrians and bicycles around the pedestrian crossing Since there are many pedestrians and bicycles on the right turn pedestrian crossing and there is little risk of overlooking them when turning right, the warning output may be limited without detecting pedestrians.

  The above pedestrian detection process is based on the assumption that the host vehicle is stopped. However, as in the case of detecting a pedestrian behind a forward parked vehicle in step 403 in FIG. In the simple difference processing as described above, a difference occurs depending on the movement of the own vehicle. Therefore, there is a problem that the movement of the pedestrian cannot be accurately captured.

  Thus, an example will be described in which an optical flow between time-series images is used for image processing for detecting a pedestrian behind a forward parked vehicle. In this method, the image area can be arbitrarily divided and the weight can be freely set, and the moving direction and moving speed of the detection target in that area can be determined, so that the pedestrian can be accurately identified even when the vehicle is moving. There is an advantage that it can be detected.

  The optical flow is a method for detecting a motion in an image by a set of vector quantities having a size and a direction. In the image processing apparatus, image data and area data are input, and an optical flow is output. The area data is represented by the position (tx, ty) and size (tw, th) of the reference area and the position (sx, sy) and size (sw, wh) of the search area as shown in FIG. It is data. In order to calculate the optical flow, at least one reference region (X coordinate tx, Y coordinate ty, width tw pixels, height th pixels) is set for a face image filtered at a certain time t, and the time The same number of search areas (X coordinate sx, Y coordinate sy, width sw pixel, height sh pixel) as the reference area set for the (t + 1) filtered face image are set. In these two areas, the reference area is set inside the search area as shown in FIG.

In the calculation of the optical flow of this embodiment, a technique called a block matching method is generally used, and an area having the highest similarity to the reference area of the previous frame is calculated from the search area of the current frame. As shown in FIGS. 16 and 17, the similarity is calculated by calculating the grayscale data in the reference area as T, an arbitrary position (xd, yd) in the search area,
-(sw-tw) / 2 〈xd 〈(sw-tw) / 2,
-(sh-th) / 2 <yd <(sh-th) / 2 (1)
Assuming that the grayscale data having the same size as the reference area cut out from the candidate area S (xd, yd), the similarity θ is defined as the angle θ formed by the reference area T and the candidate area S by the following equation.
cos θ (xd, yd) = T · S (xd, yd) / (| T | · | S (xd, yd) |) (2)

  The similarity θ between the reference area and the candidate area is obtained at all positions in the search area. The coordinate (xd, yd) that maximizes the similarity θ is taken as the motion vector (xd, yd) of the reference region. In addition to the method described in this embodiment, the optical flow (motion detection) calculation method is supervised by Nobuyuki Yagi “Digital Video Processing” (Edition of the Video Information Media Society, pp129-139, 2000, published by Ohm) A number of techniques for detecting motion from moving images have been introduced in the literature, and can be used.

  FIG. 18 shows an example in which a pedestrian existing behind a parked vehicle is captured by a motion vector of an optical flow when approaching a parked vehicle in front. A pedestrian's approach can be detected by detecting a motion vector generated in the right direction at the foot part of the pedestrian from radial motion vectors generated from an image vanishing point due to the movement of the host vehicle.

<< Modification of Embodiment of Invention >>
In the above-described embodiment, the driver recognizes the parked vehicle in front of the vehicle, operates the parked vehicle confirmation switch 211, and detects the pedestrian hidden behind the parked vehicle. Next, a modified example in which recognition of a parked vehicle in front of the vehicle is performed automatically will be described.

  FIG. 19 shows a configuration of a modified example. In FIG. 19, the devices 2001 to 2010 are the same as the devices 201 to 210 shown in FIG. FIG. 20 is a flowchart showing a process for detecting a pedestrian behind a parked vehicle. Also in FIG. 20, the processing of 2101, 2103 to 2105 is the same as the processing of 401 and 403 to 405 in FIG. In this modification, a radar 2011 is used instead of the parked vehicle confirmation switch 211 shown in FIG. The radar 2011 is installed in the front part of the vehicle, scans the front of the vehicle, recognizes an obstacle, and detects the direction and distance of the obstacle. In step 2102 of FIG. 20, if an obstacle is detected on the traveling road by the radar 2011, it is recognized as a parked vehicle on the traveling road, and the process proceeds to step 2103 to be a pedestrian behind the front parked vehicle. The detection process is performed.

  In addition, you may make it recognize the parked vehicle ahead of a vehicle by processing the image ahead of the vehicle imaged with the camera.

  As described above, according to the embodiment and the modification thereof, the current position and the running state of the vehicle are detected, the imaging range around the vehicle is changed according to the current position and the running state, and the time-series images are taken. Since the pedestrian approaching the vehicle is detected and alarmed, the imaging range is accurately determined even if the driver's blind spot changes depending on the current position of the vehicle and the driving state. The pedestrian in the blind spot range can be reliably detected and alarmed.

  The correspondence between the constituent elements of the claims and the constituent elements of the embodiment is as follows. That is, the cameras 201 and 2001 are imaging means, the navigation devices 206 and 2006 are vehicle position detection means, the vehicle speed sensors 207 and 2007 and the turn signal switches 209 and 2009 are running state detection means, the motors 204 and 2004, and the motor control device. 205 and 2005 and the control devices 203 and 2003 constitute imaging range variable means, the image processing devices 202 and 2002 and the control devices 203 and 2003 constitute pedestrian detection means, and the warning devices 210 and 2010 constitute alarm means. In addition, unless the characteristic function of this invention is impaired, each component is not limited to the said structure.

  In the above-described embodiment, an example in which the camera is installed at the lower left and right front ends of the vehicle has been described. However, the installation location of the camera is not limited to the embodiment. Further, the number of cameras and the imaging range of each camera are not limited to the above-described embodiment.

It is a figure which shows the structure of one embodiment. It is a flowchart which shows the pedestrian warning program of one Embodiment. It is a flowchart which shows the program which detects and alert | reports the pedestrian who exists behind the front parking / stopping vehicle. It is a figure which shows the attachment position of a camera. It is a figure which shows the imaging range of a camera. It is a figure which shows the example which installed the near-infrared lamp in the vicinity of the camera. It is a flowchart which shows the detection process of a passing vehicle. It is a figure which shows the general driver | operator's viewing angle. It is a figure which shows a driver | operator's viewing angle and the imaging range of a camera when waiting for a right turn, seeing the space of the passing vehicle which opposes at a T-shaped intersection. It is a figure which shows the time series image of the camera which caught the passing vehicle. It is a figure which shows the difference generation | occurrence | production situation of the camera image which caught the passing vehicle. It is a flowchart which shows the subroutine for detecting the pedestrian approaching a right turn destination pedestrian crossing. It is a figure which shows the time series image of the camera which captured the pedestrian approaching a right turn destination pedestrian crossing. It is a figure which shows the imaging range of the camera at the time of passing through a no-signal intersection. It is a figure for demonstrating the method to detect the pedestrian who exists in the shadow of the front parking / stopping vehicle. It is a figure for demonstrating the application method of an optical flow. It is a figure for demonstrating the application method of an optical flow. It is a figure for demonstrating the application method of an optical flow. It is a figure which shows the structure of the modification of one Embodiment. It is a flowchart which shows the program which detects and alert | reports the pedestrian who exists behind the front parking and stopping vehicle of the modification of one Embodiment.

Explanation of symbols

201, 2001 Camera 202, 2002 Image processing device 203, 2003 Control device 204, 2004 Motor 205, 2005 Motor control device 206, 2006 Navigation device 207, 2007 Vehicle speed sensor 208, 2008 Turn signal switch 209, 2009 Brake switch 210, 2010 Warning Device 211 Parked vehicle confirmation switch 2011 Radar

Claims (7)

Imaging means for imaging the surroundings of the vehicle;
Vehicle position detection means for detecting the current position of the vehicle;
Traveling state detecting means for detecting the traveling state of the vehicle;
An imaging range variable means for changing the imaging range of the imaging means according to the current position of the vehicle detected by the vehicle position detection means and the running state of the vehicle detected by the running state detection means;
A pedestrian detection means for detecting a pedestrian approaching the vehicle by processing a time-series image captured by the imaging means;
A pedestrian warning device comprising: warning means for warning when a pedestrian approaching the vehicle is detected by the pedestrian detection means. In the pedestrian warning device according to claim 1,
When the vehicle position detection means detects that the vehicle is in the vicinity of the center of the intersection, and the traveling state detection means detects the operating state of the right turn side turn signal switch at a state below a predetermined vehicle speed, the imaging range variable means A pedestrian warning device characterized in that the imaging range of the imaging means is the range on the right rear side of the vehicle. In the pedestrian warning device according to claim 1,
When the vehicle is located at a no-signal intersection by the vehicle position detection means and the state below a predetermined vehicle speed is detected by the traveling state detection means, the imaging range of the imaging means is set to the left and right sides of the vehicle by the imaging range variable means. A pedestrian warning device characterized by In the pedestrian warning device according to claim 1,
A pedestrian warning device characterized in that when the vehicle position detecting means detects that the vehicle is not approaching an intersection, the imaging range variable means sets the imaging range of the imaging means to the front of the vehicle. In the pedestrian warning device according to any one of claims 1 to 4,
Installing the imaging means at the bottom of the vehicle;
The pedestrian detection means is characterized by processing a time-series image captured by the imaging means to detect a movement of a pedestrian's foot seen from the bottom of another vehicle and determining whether or not the vehicle approaches the vehicle. Pedestrian alarm device to do. In the pedestrian warning device according to any one of claims 1 to 5,
The pedestrian detection device is a pedestrian alarm device characterized in that the time-series image captured by the imaging device is processed by difference calculation to detect a pedestrian approaching the vehicle. In the pedestrian warning device according to any one of claims 1 to 5,
The pedestrian detection device is a pedestrian warning device characterized by detecting a pedestrian approaching a vehicle by processing a time-series image captured by the imaging device by an optical flow calculation.

Images