JP2017207920A - Reverse travelling vehicle detection device and reverse travelling vehicle detection method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technique that can more efficiently detect a reverse travelling vehicle.SOLUTION: A likelihood calculation unit is configured to: as to each forward vehicle detected by a vehicle detection unit, execute characteristic recognition processing of recognizing a characteristic image characterizing that the forward vehicle is a reverse travelling vehicle from an image to be obtained from an imaging device of an own vehicle; calculate a likelihood indicative of assurance in which the characteristic image can be recognized, in which a plurality of pieces of characteristic recognition processing are prepared by the number of pieces of characteristic recognition processing in accordance with a category of the characteristic image; as to the forward vehicle, execute the plurality of pieces of characteristic recognition processing one by one in order from the characteristic recognition processing small in amount of calculation; and obtain a total value of the likelihood calculated by the execution of respective characteristic recognition processing. A determination unit is configured to: determine the forward vehicle which is determined that the total value of the likelihood exceeds a threshold as the reverse travelling vehicle; and suspend processing of the likelihood calculation unit as to the forward vehicle determined as the reverse travelling vehicle.SELECTED DRAWING: Figure 6

Description

  The present disclosure relates to a technique for detecting a reverse traveling vehicle.

Conventionally, there has been known a technique for detecting another vehicle that runs backward on a road on the lane in which the host vehicle is traveling as a backward running vehicle and informing the driver of the own vehicle of the presence of the detected backward running vehicle. Yes.
For example, in Patent Document 1, when a headlamp of another vehicle is recognized from an image captured by a front camera mounted on the host vehicle, and the lane in which the other vehicle is traveling matches the lane in which the host vehicle is traveling. A reverse running vehicle detection device that detects other vehicles as reverse running vehicles when determined is proposed.

JP 2011-59910 A

  Specifically, in the reverse vehicle detection device of Patent Document 1, in image processing for recognizing a headlamp of another vehicle, predetermined conditions such as the number and arrangement are satisfied from among the light sources specified based on the luminance of the image. A white light source is extracted, and the extracted light source is recognized as a headlamp. As described above, the reverse running vehicle detection device disclosed in Patent Document 1 detects the reverse running vehicle on the assumption that the headlight of the preceding vehicle is illuminated at night or the like.

  On the other hand, even when the headlight is not illuminated in the daytime, etc., by recognizing each part including the headlamp of the other vehicle from the image, the other vehicle facing the host vehicle side in front A method of detecting as a reverse running vehicle is conceivable. However, in such a reverse vehicle detection method, since the shape and arrangement of each part are relatively different depending on the vehicle type, the calculation amount of processing relating to image recognition increases. For this reason, if only such a reverse running vehicle detection method is adopted, there is a possibility that the timing for notifying the passenger of the own vehicle of the presence of the reverse running vehicle in the daytime or the like may be delayed.

  The present disclosure provides a technique capable of detecting a reverse vehicle more efficiently.

  The reverse running vehicle detection device of the present disclosure is a device that detects a reverse running vehicle, and includes a vehicle detection unit, a likelihood calculation unit, a determination unit, and a notification unit. The reverse running vehicle in the present disclosure refers to a lane in which the host vehicle is traveling (hereinafter referred to as a traveling lane), or a separate lane that can be changed from the traveling lane and a traveling lane. The other vehicle that runs backward on the road shown (hereinafter referred to as the target road).

  A vehicle detection part performs the process which detects the other vehicle (henceforth front vehicle) which exists on the object road ahead of the own vehicle. The likelihood calculation unit indicates that the front vehicle is a reverse running vehicle from an image obtained by an imaging device mounted on the host vehicle so as to capture the front of the host vehicle for each front vehicle detected by the vehicle detection unit. Processing for recognizing the feature image to be characterized (hereinafter, feature recognition processing) is executed. In addition, the likelihood calculation unit performs a process of calculating a likelihood indicating the probability that the feature image can be recognized by executing the feature recognition process.

  The determination unit determines whether or not the total value of the likelihoods calculated by the likelihood calculation unit exceeds a predetermined threshold value, and determines that the preceding vehicle determined that the total likelihood value exceeds the threshold value is a reverse vehicle Processing to determine is performed. The notification unit performs a process of notifying at least a passenger of the own vehicle of the presence of the reverse running vehicle determined by the determination unit.

  In the reverse vehicle detection device according to the present disclosure, a plurality of feature recognition processes are prepared according to the type of feature image. In addition, the likelihood calculation unit executes a plurality of feature recognition processes for each forward vehicle one by one in order from the feature recognition process with a small calculation amount, and calculates the total likelihood calculated by executing each feature recognition process. Configured to determine a value. Furthermore, the determination unit is configured to stop the processing of the likelihood calculation unit for the preceding vehicle that is determined that the total likelihood value exceeds the threshold value.

  Here, among the plurality of feature recognition processes, for example, the feature recognition process for recognizing an image (hereinafter referred to as a vehicle front image) indicating that the preceding vehicle is facing the host vehicle side as a feature image in the daytime or the like is described above. As described above, the amount of calculation for the image recognition process is relatively large.

  On the other hand, in the reverse running vehicle detection device of the present disclosure, for example, in the daytime or the like, among a plurality of feature recognition processes that differ depending on the type of the feature image, the calculation amount is larger than the image recognition that recognizes the vehicle front image as the feature image. It is possible to execute the feature recognition process one by one in order from the process of recognizing another feature image with few.

  Further, in the reverse running vehicle detection device of the present disclosure, by executing each feature recognition process, a total value of likelihood indicating the likelihood that the feature image can be recognized is obtained, and when the total value of likelihood exceeds the threshold value At the time of the determination, the forward vehicle determined in this way is determined as a reverse traveling vehicle. Further, when it is determined that the total likelihood value exceeds the threshold value, the execution of the feature recognition process with a larger calculation amount for the preceding vehicle is canceled. For this reason, it is possible to detect a reverse vehicle without necessarily performing feature recognition processing with a large calculation amount.

Therefore, according to the reverse running vehicle detection device of the present disclosure, the reverse running vehicle can be detected more efficiently.
Further, according to the reverse running vehicle detection method of the present disclosure, the same effect as that already described in the reverse running vehicle detection device of the present disclosure can be obtained for the same reason as described above.

It is a block diagram which shows the structure of the reverse running vehicle detection apparatus. 3 is a block diagram showing a functional configuration of a control unit 10. FIG. It is a flowchart of a road determination process. It is a flowchart of a reverse vehicle determination process. It is a flowchart of the alerting | reporting process regarding an outward warning. It is explanatory drawing of a feature image.

Hereinafter, embodiments for carrying out the invention will be described with reference to the drawings.
[1. First Embodiment]
[1-1. Constitution]
A reverse running vehicle detection device 1 shown in FIG. 1 is a device mounted on a host vehicle so as to detect a reverse running vehicle, and includes a front camera 2, a millimeter wave radar 3, a navigation device 4, and inter-vehicle communication. A unit 5, a headlamp ECU 6, a horn ECU 7, a vehicle speed sensor 8, and a control unit 10 are provided.

  A reverse running vehicle is composed of a lane in which the host vehicle is traveling (hereinafter referred to as a traveling lane), or, if there is another lane that can be changed from the traveling lane, the other lane and the traveling lane. The other vehicle that runs backward on the road (hereinafter referred to as the target road). The target roads composed of the above-mentioned traveling lanes include one-way roads, roads that are not one-way but have a relatively small road width, roads for one lane in general one-lane roads, etc. Is included. The target roads composed of the above-mentioned separate lanes and traveling lanes include roads for two lanes in the same traveling direction on roads on one lane, roads for three lanes in the same traveling direction on roads on one lane, etc. Is included. In other words, the target road includes a road on a different lane whose traveling direction is the same as the travel lane, in addition to a road on the travel lane.

  The front camera 2 corresponds to an imaging device mounted on the host vehicle so as to capture the front of the host vehicle, and is configured to output a captured image (hereinafter “front image”) to the control unit 10. The The front camera 2 is fixed, for example, by adhesion to the upper end of the front windshield in the passenger compartment so that the lens is not exposed to wind and rain and does not interfere with the driver's view. The front camera 2 has a known image sensor such as a CMOS or a CCD, for example, and takes an image of an area extending in a predetermined range having an optical axis slightly downward toward the front of the host vehicle. Specifically, in the front camera 2, light incident from the front of the host vehicle is photoelectrically converted by the imaging device, a signal read out as an accumulated charge voltage is amplified, and a predetermined luminance is obtained by A / D conversion. It is converted into a gradation digital image (ie, a front image).

  The millimeter wave radar 3 is mounted on the host vehicle so as to detect the position and relative speed of an object existing in front of the host vehicle, and information (hereinafter referred to as the position and relative speed of the detected object). Radar information) is output to the control unit 10. For example, the millimeter wave radar is installed in the front grill of the host vehicle, and transmits a millimeter wave as a radar wave while scanning a predetermined angle range in the left-right direction centering on the front direction of the host vehicle. Receive the reflected wave. Specifically, in a millimeter wave radar, a transmission wave modulated with a triangular wave is output from an antenna, and a beat signal is acquired by receiving and mixing a reflected wave reflected from an object ahead of the host vehicle with the antenna. . Since the waveform of the beat signal changes due to interference generated according to the distance to the object and the relative speed, the relative distance and the relative speed are calculated from the waveform. In addition, since the reflected wave is received if there is an object in the irradiation direction of the transmission wave, it is possible to detect the direction of the object existing in front of the host vehicle.

  The navigation device 4 is a well-known device that displays the current position of the host vehicle together with a road map or provides route guidance to a destination. Information on the target road (hereinafter, target road information) is sent to the control unit 10. Configured to output. Specifically, the navigation device 4 detects the current position of the host vehicle using the arrival time of the radio wave received from the satellite of the global navigation satellite system (that is, GNSS), and further detects the speed and yaw rate of the host vehicle. Use to correct the current position. The navigation device 4 has a map database including road map information associated with position information such as latitude and longitude. The road map information is a table-like database in which link information related to the links constituting the road is associated with node information related to the nodes connecting the links. In addition to the link length and connection node, the link information includes information such as the width of the road, the number of lanes constituting the road, the traveling direction of the road, and the presence or absence of one-way traffic. For this reason, the navigation device 4 can identify the target road from the road map information based on the position of the host vehicle, and can acquire link information regarding the identified target road as the target road information. In addition, the navigation device 4 uses road-to-vehicle communication, etc., when acquiring information such as the number of lanes temporarily decreasing due to construction, etc., or information where the presence or absence of one-way traffic has been changed, Such information may be added to the target road information and output to the control unit 10. Further, a well-known locator having the same function may be used in place of the navigation device 4 except that the display function and the route guidance function are removed from the navigation device 4.

  The vehicle speed sensor 8 is a known sensor that detects the speed of the host vehicle (that is, the host vehicle speed), and is configured to output information indicating the host vehicle speed (hereinafter referred to as vehicle speed information) to the control unit 10, the navigation device 4, and the like. Is done.

  The inter-vehicle communication unit 5 is mounted on the host vehicle so as to transmit and receive various types of information to and from other vehicles by wireless communication. The vehicle-to-vehicle communication unit 5 includes an occupant including a driver of a reverse running vehicle in accordance with a command from the control unit 10. The wireless information indicating the warning for notifying that the vehicle is traveling backward is transmitted to the vehicle traveling backward. The inter-vehicle communication unit 5 further wirelessly transmits to the other vehicle wireless information including a warning for notifying the occupant of the vehicle other than the reverse vehicle around the host vehicle of the presence of the reverse vehicle. It may be configured as follows. In the following, a warning for notifying the occupant including the driver of the reverse running vehicle that the vehicle is running backward, and the presence of the reverse running vehicle are notified to passengers of other vehicles other than the reverse running vehicle around the host vehicle. This warning is collectively referred to as an outward warning.

  The headlamp ECU 6 is an electronic control unit (that is, ECU) that controls the headlamp of the host vehicle, and is configured to issue an outward warning by a headlight in accordance with a command from the control unit 10. The outward warning here is realized by, for example, passing one or more times for temporarily irradiating a high beam headlight.

  The horn ECU 7 is an ECU that controls the horn of the host vehicle, and is configured to issue an outward warning by the horn in accordance with a command from the control unit 10. The outward warning here is realized, for example, by outputting a strong horn sound one or more times.

  The control unit 10 is configured around a well-known microcomputer having a CPU 11 and a semiconductor memory (hereinafter, memory 12) such as a RAM, a ROM, and a flash memory. Various functions of the control unit 10 are realized by the CPU 11 executing a program stored in a non-transitional physical recording medium. In this example, the memory 12 corresponds to a non-transitional tangible recording medium that stores a program. Further, by executing this program, a method corresponding to the program is executed. The number of microcomputers constituting the control unit 10 may be one or plural.

  As shown in FIG. 2, the control unit 10 includes a vehicle detection unit 20, a speed detection unit 25, a likelihood calculation unit 30, and a determination unit 40 as a functional configuration realized by the CPU 11 executing a program. And a notification unit 50. The technique for realizing these elements constituting the control unit 10 is not limited to software, and some or all of the elements may be realized using hardware that combines a logic circuit, an analog circuit, and the like.

  The vehicle detection unit 20 is configured to perform a process of detecting another vehicle (hereinafter referred to as a forward vehicle) present on the target road in front of the host vehicle. Specifically, the vehicle detection unit 20 detects an object on the image from the front image output by the front camera 2 based on information such as luminance and color, and among the detected objects, An image whose contour represents the contour of the vehicle (hereinafter, vehicle image) is recognized. The vehicle image recognition is performed by, for example, matching processing using a previously registered object model for vehicle image recognition. The object model for vehicle image recognition shows the general characteristics of the vehicle to such an extent that the vehicle can be distinguished from those other than vehicles such as pedestrians and bicycles. For this reason, as shown in FIG. 6, the vehicle detection unit 20 specifies the general characteristics of the vehicle for the recognized vehicle image, but does not specify, for example, whether the vehicle is facing forward.

  In addition, the vehicle detection unit 20 recognizes one or a plurality of lanes including a traveling lane by detecting a lane boundary line such as a white line from the front image output by the front camera 2, Association with the lane indicated by the target road information output by the navigation device 4 is performed. Then, when the lane in which the other vehicle indicated by the vehicle image is traveling can be associated with the lane indicated by the target road information, the other vehicle exists on the target road in front of the host vehicle ( That is, it is determined that the vehicle is ahead. The vehicle detection unit 20 is configured to output the vehicle image of the preceding vehicle detected in this way to the likelihood calculation unit 30.

  Furthermore, the vehicle detection unit 20 detects the position of the front vehicle having the road surface in the real space as a two-dimensional plane based on the position of the front vehicle and the position of the infinity point (that is, FOE) in the front image. Information relating to the detected position of the preceding vehicle (hereinafter referred to as position detection information) is output to the speed detection unit 25.

  The speed detection unit 25 is configured to perform a process of detecting a speed at which the preceding vehicle detected by the vehicle detection unit 20 travels toward the own vehicle (hereinafter referred to as an approach speed). Specifically, the speed detection unit 25 specifies the relative speed of the front vehicle at each position based on the position detection information output by the vehicle detection unit 20 and the radar information output by the millimeter wave radar 3. To do. Then, the speed detection unit 25 calculates the approach speed of the front vehicle at each position based on the relative speed thus specified and the host vehicle speed indicated by the vehicle speed information output by the vehicle speed sensor 8. The speed detection unit 25 can add the information on the approaching speed of the preceding vehicle calculated in this way to the position detection information and output it to the likelihood calculation unit 30.

  The likelihood calculating unit 30 recognizes a feature image that characterizes that the preceding vehicle is a reverse vehicle from the front image obtained by the front camera 2 for each preceding vehicle detected by the vehicle detecting unit 20 (hereinafter referred to as a feature). Recognition process). And the likelihood calculation part 30 is comprised so that the process which calculates the likelihood which shows the probability that the feature image was able to be recognized by execution of a feature recognition process will be performed. Specifically, the likelihood calculation unit 30 performs a plurality of feature recognition processes for each preceding vehicle one by one in order from a feature recognition process with a small amount of calculation, and is calculated by executing each feature recognition process. It is configured to determine a total likelihood value.

  In the present embodiment, the plurality of feature recognition processes include a process for recognizing a tail lamp non-existent image, a process for recognizing a face image, and a process for recognizing a vehicle front image, respectively, in ascending order of calculation amount. . Thus, a plurality of feature recognition processes are prepared in a number corresponding to the type of feature image.

  As shown in FIG. 6, the tail lamp non-existing image is an image indicating that the tail lamp of the preceding vehicle does not face the own vehicle side in the vehicle image output by the vehicle detection unit 20. The face image is an image indicating that the face of the occupant of the vehicle ahead is facing the host vehicle. The vehicle front image is an image indicating that the preceding vehicle is facing the host vehicle. Recognition of these feature images is performed, for example, by matching processing using object models for each feature image registered in advance.

  The recognition of the tail lamp non-existent image is a process of detecting a feature that the red light source or the brown region does not exist symmetrically from the vehicle image output by the vehicle detection unit 20 based on information such as luminance and color. It is realized by doing. In other words, the likelihood calculation unit 30 determines the likelihood value related to the recognition of the tail lamp non-existing image when the characteristic of the tail lamp that the red light source or the brown area exists symmetrically cannot be detected in this process. Set to the highest value. In addition, the likelihood calculating unit 30, for example, when it is not possible to detect a red light source at night or the like, when it is not possible to detect a brown region at daytime, or when it is not possible to detect that the light source exists symmetrically, A predetermined likelihood value is set according to each case. In this way, the likelihood calculating unit 30 is configured to calculate a likelihood value related to recognition of a tail lamp non-present image.

  The recognition of the face image uses the well-known face recognition algorithm from the vehicle image output by the vehicle detection unit 20 to determine the facial features such as the relative position and size of the facial parts, the shape of the eyes, nose, cheekbones and jaws. This is realized by performing a process of searching for at least one of a plurality of conditions shown. In this process, the likelihood calculating unit 30 sets the likelihood value related to the recognition of the face image to the highest value when, for example, an image region that satisfies all the conditions indicating the facial features can be detected. . The likelihood value is determined in advance for each type of the condition. Therefore, the likelihood calculating unit 30 can set the likelihood value to a high value, for example, according to the number of conditions satisfied among a plurality of conditions indicating facial features. In this way, the likelihood calculating unit 30 is configured to calculate a likelihood value related to recognition of a face image.

  The recognition of the vehicle front image has the same basic principle as the recognition of the face image. From the vehicle image output by the vehicle detection unit 20, the relative position and size of the parts arranged on the front side of the vehicle, the front This is realized by performing a process of searching for at least one of a plurality of conditions indicating the front characteristics of the vehicle such as the shape of the glass, the front grille, and the headlamp. In this process, the likelihood calculating unit 30 sets the likelihood value related to the recognition of the vehicle front image to the highest value when, for example, an image region that satisfies all the conditions indicating the front characteristics of the vehicle can be detected. Set. The likelihood value is determined in advance for each type of the condition. For this reason, the likelihood calculation part 30 can set the value of likelihood to a high value according to the number of the conditions satisfy | filled among the some conditions which show the characteristic of a vehicle front, for example. In this way, the likelihood calculating unit 30 is configured to calculate the likelihood value related to the recognition of the vehicle front image.

  In addition, the likelihood calculating unit 30 is configured to set a converted value that becomes a large value according to the approach speed detected by the speed detecting unit 25 and to add the set converted value to the total likelihood value. . Specifically, the likelihood calculating unit 30 identifies the approaching speed of the preceding vehicle that is the target of the feature recognition process being performed based on the position detection information output by the speed detecting unit 25, and identifies the identified preceding vehicle. The conversion value is set so that the likelihood value becomes higher as the approach speed becomes higher. The coefficient for converting the approach speed to the likelihood value (that is, the converted value) may be a fixed value or a variable value that changes linearly or nonlinearly according to the approach speed. .

  The determination unit 40 determines whether or not the total likelihood value calculated by the likelihood calculation unit 30 exceeds a predetermined threshold value, and reverses the preceding vehicle that has been determined that the total likelihood value exceeds the threshold value. It is comprised so that the process (henceforth a reverse running vehicle determination process) determined as a running vehicle may be performed. Further, the determination unit 40 is a road whose road width is smaller than a predetermined reference width for the target road, and the target road is not a one-way road, the vehicle detection unit 20 and the likelihood calculation unit 30. Is configured to perform a process of stopping at least one of the processes (hereinafter, road determination process). Each process executed by the determination unit 40 will be described later.

  The notification unit 50 is configured to perform a process of notifying at least the occupant of the host vehicle of the presence of the reverse running vehicle determined by the determination unit 40. Thus, the warning for notifying the passenger of the own vehicle of the presence of the reverse running vehicle is referred to as an inward warning. The inward warning is given via an audio output device or a display device mounted on the host vehicle. For example, a liquid crystal display, a head-up display, or the like is used as the display device. In an inward warning via the display device, for example, the front vehicle on the front image displayed on the liquid crystal display is highlighted so as to be surrounded by a frame, or the vehicle is viewed through the front windshield viewed from the driver of the own vehicle. The vehicle ahead can be highlighted using a head-up display so as to surround the vehicle with a frame.

  Further, in the present embodiment, the notification unit 50 further communicates between the vehicle, the headlights of the host vehicle, and the horn to the occupant of the reverse traveling vehicle determined by the determination unit 40 as the processing related to the outward warning described above. It is comprised so that the process which alert | reports using at least one may be performed. The notification process regarding the outward warning executed by the notification unit 50 will be described later.

[1-2. processing]
[1-2-1. Road judgment process]
First, the road determination process executed by the determination unit 40 will be described with reference to the flowchart of FIG.

  When this process is started, the determination unit 40 first acquires target road information regarding the target road where the preceding vehicle detected by the vehicle detection unit 20 exists from the vehicle detection unit 20 or the navigation device 4 in S110. . In addition, when there are a plurality of forward vehicles detected by the vehicle detection unit 20, target road information corresponding to each of the plurality of forward vehicles may be acquired, or a target corresponding to any one of the forward vehicles. Road information may be acquired. And the determination part 40 implements the process of following S120-S150 about the acquired object road information.

  In S120, the determination unit 40 determines whether the road width of the target road is smaller than a predetermined reference width for the target road based on the target road information acquired in S110. Then, when it is determined that the target road is a road whose road width is smaller than the reference width, the process proceeds to S130, and when it is determined that the road width is equal to or larger than the reference width, the process proceeds to S150. To do. Here, the reference width indicates, for example, the width of a general two-lane road. For this reason, roads whose road width is smaller than the reference width include roads of one lane, and roads whose road width is greater than the reference width include roads of three lanes or three lanes. A road with a size of minutes is included. Therefore, such determination regarding the road width of the target road can be made based on, for example, the number of lanes constituting the target road among the information included in the target road information.

  In subsequent S130, the determination unit 40 determines whether or not the target road is a one-way road based on the target road information acquired in S110. If it is determined that the target road is a one-way road, the process proceeds to S140. If it is determined that the target road is not a one-way road, the process proceeds to S150.

  In S140, the determination unit 40 stops the processes of the vehicle detection unit 20 and the likelihood calculation unit 30, and ends the road determination process. On the other hand, in S150, when the processes of the vehicle detection unit 20 and the likelihood calculation unit 30 are being executed, the determination unit 40 continues the execution of these processes, and the vehicle detection unit 20 and the likelihood calculation unit 30. If these processes are stopped, the execution of these processes is resumed, and the road determination process is terminated. The road determination process is resumed at a predetermined timing after the end.

  As described above, the determination unit 40 satisfies the condition that the target road is a road whose width is smaller than the reference width and the target road is not a one-way road (hereinafter referred to as a road condition). The detection unit 20 and the likelihood calculation unit 30 are configured to stop processing. And the determination part 40 is comprised so that execution of the process of the vehicle detection part 20 and the likelihood calculation part 30 may be continued or restarted, when such road conditions are not satisfy | filled.

  Note that when such a road condition is satisfied, the determination unit 40 may stop the processing of the likelihood calculation unit 30 by stopping the processing of the vehicle detection unit 20, or the vehicle detection unit 20 The process of the likelihood calculating unit 30 may be stopped without stopping the process. The determination unit 40 may further stop the processing of the speed detection unit 25 when such a road condition is satisfied. Further, even when such a road condition is satisfied, the determination unit 40, when an approaching speed equal to or higher than a predetermined speed is detected by the speed detection unit 25 for the preceding vehicle, the vehicle detection unit 20 and the likelihood. You may make it continue or restart execution of the process of the calculation part 30. FIG.

[1-2-2. Reverse running vehicle judgment process]
Next, the reverse vehicle determination process executed by the determination unit 40 will be described with reference to the flowchart of FIG.

  When the present process is started, the determination unit 40 first determines whether or not one or more forward vehicles are detected by the vehicle detection unit 20 in S210. If it is determined that one or more forward vehicles have been detected, the process proceeds to S220. If it is determined that no forward vehicle has been detected, the reverse vehicle determination process is terminated. Note that the reverse vehicle determination process is resumed at a predetermined timing after completion.

  In S <b> 220, the determination unit 40 extracts one target vehicle from one or more forward vehicles detected by the vehicle detection unit 20. For example, as the target vehicle, one forward vehicle is extracted from the one or a plurality of forward vehicles in order of close position from the host vehicle or in descending order of approach speed.

  In subsequent S <b> 230, the determination unit 40 acquires the converted value from the likelihood calculating unit 30 by outputting a command for calculating a converted value related to the approach speed of the target vehicle to the likelihood calculating unit 30. As described above, the converted value is a value obtained by converting the approach speed into a likelihood value related to image recognition, and becomes a large value according to the approach speed.

  In subsequent S240, the determination unit 40 determines whether or not the converted value acquired in S230 (that is, the likelihood value P0) exceeds a preset likelihood threshold value PD. If it is determined that the likelihood value P0 exceeds the threshold value PD, the process proceeds to S310. If it is determined that the likelihood value P0 is equal to or less than the threshold value PD, the process proceeds to S250. The likelihood threshold PD may be a fixed value or a variable value. The likelihood threshold PD is set to a larger value as the approach speed is smaller than the own vehicle speed based on the approach speed of the target vehicle and the own vehicle speed (that is, the own vehicle speed), for example. The larger the vehicle speed, the smaller the value can be set.

  In S250, the determination unit 40 outputs, as the first feature recognition, a command for calculating the likelihood value P1 related to the recognition of the tail lamp non-existing image for the target vehicle to the likelihood calculation unit 30. The total likelihood value including the value P <b> 1 is acquired from the likelihood calculating unit 30. As described above, the likelihood value P1 is a value indicating the probability that an image indicating that the tail lamp of the target vehicle is not facing the host vehicle side in the vehicle image can be recognized. The value is so large that is not detected.

  In subsequent S260, the determination unit 40 determines whether or not the total likelihood value P0 + P1 acquired in S250 exceeds a likelihood threshold value PD. When it is determined that the total likelihood value P0 + P1 exceeds the threshold value PD, the process proceeds to S310, and when it is determined that the total likelihood value P0 + P1 is equal to or less than the threshold value PD, the process proceeds to S270.

  In S270, the determination unit 40 outputs, as the second feature recognition, a command for calculating the likelihood value P2 related to the recognition of the face image of the target vehicle to the likelihood calculation unit 30, thereby this value P2 Is obtained from the likelihood calculation unit 30. As described above, the likelihood value P2 is a value indicating the likelihood that an image indicating that the occupant (for example, the driver) of the target vehicle is facing the host vehicle side in the vehicle image can be recognized. As the facial feature is detected from the image, the value increases.

  In subsequent S280, the determination unit 40 determines whether or not the total likelihood value P0 + P1 + P2 acquired in S270 exceeds the likelihood threshold value PD. When it is determined that the total likelihood value P0 + P1 + P2 exceeds the threshold value PD, the process proceeds to S310, and when it is determined that the total likelihood value P0 + P1 + P2 is equal to or less than the threshold value PD, the process proceeds to S290.

  In S290, the determination unit 40 outputs a command for calculating the likelihood value P3 related to the recognition of the vehicle front image for the target vehicle to the likelihood calculation unit 30 as the third feature recognition, thereby obtaining this value. The total likelihood value including P3 is acquired from the likelihood calculating unit 30. As described above, the likelihood value P3 is a value indicating the probability that an image indicating that the target vehicle is facing the host vehicle side in the vehicle image can be recognized. The larger the value is, the larger the value is detected.

  In subsequent S300, the determination unit 40 determines whether or not the total likelihood value P0 + P1 + P2 + P3 acquired in S290 exceeds a likelihood threshold value PD. When it is determined that the total likelihood value P0 + P1 + P2 + P3 exceeds the threshold value PD, the process proceeds to S310, and when it is determined that the total likelihood value P0 + P1 + P2 + P3 is equal to or less than the threshold value PD, the process proceeds to S320.

  In S <b> 310, the determination unit 40 determines that the target vehicle is a reverse running vehicle, and outputs a command for performing notification processing regarding the inward warning and the outward warning about the reverse running vehicle to the notification unit 50. Then, the reverse vehicle determination process is terminated. As described above, the inward warning is a warning given to the passenger of the host vehicle in order to notify the presence of the reverse running vehicle. An outward warning is a warning given to a reverse-traveling vehicle occupant in order to notify that the vehicle is running in reverse, or to other vehicle occupants around the host vehicle in order to notify the presence of the reverse-traveling vehicle. This is a warning.

  In S320, the determination unit 40 determines whether or not the extraction as the target vehicle in S220 has been completed for all the preceding vehicles among the one or more preceding vehicles determined to have been detected in S210. If it is determined that all the preceding vehicles have been completed, the reverse running vehicle determination process is terminated. If it is determined that there is an unextracted preceding vehicle, the process returns to S220.

  As described above, the determination unit 40 determines whether or not the total likelihood value calculated by the likelihood calculation unit 30 exceeds the threshold value, and reversely runs the vehicle ahead that has determined that the total likelihood value exceeds the threshold value. It is configured to determine as a car. Moreover, the determination part 40 is comprised so that the process of the likelihood calculation part 30 about the front vehicle determined that the total value of likelihood exceeds a threshold value.

[1-2-3. Notification processing]
Finally, of the notification processing executed by the notification unit 50, notification processing related to an outward warning will be described with reference to the flowchart of FIG. This process is started when a command is output from the determination unit 40 in S310.

  When this process is started, the notification unit 50 first outputs a command for transmitting wireless information indicating an outward warning to the reverse running vehicle to the inter-vehicle communication unit 5 in S410. When this command is output, the inter-vehicle communication unit 5 outputs information related to the success or failure of the wireless transmission to the notification unit 50.

  In continuing S420, the alerting | reporting part 50 judges the presence or absence of a communication error based on the information output from the inter-vehicle communication part 5 in S410. The communication error includes, for example, a case where the inter-vehicle communication unit 5 fails to transmit the above-described wireless information, a case where a response signal for the above-described wireless information is not wirelessly transmitted from the reverse traveling vehicle to the inter-vehicle communication unit 5, etc. There is. If it is determined that there is a communication error, the process proceeds to S430, and if it is determined that communication is normally performed, the notification process regarding the outward warning is terminated.

  In S430, the notification unit 50 outputs a command for performing an outward warning by the headlight to the headlamp ECU 6. When this command is output, the headlamp ECU 6 outputs information indicating that the passing has been performed a predetermined number of times to the notification unit 50.

  In continuing S440, the alerting | reporting part 50 judges whether the behavior of the reverse running vehicle changed after information was output from the alerting | reporting part 50 in S430 based on the front image by which the reverse running vehicle is imaged. Examples of the behavior change include a case where the reverse running vehicle has moved to the shoulder side, and a case where the running reverse running vehicle has stopped. If it is determined that the behavior of the reverse traveling vehicle has changed, the notification process regarding the outward warning is terminated, and if it is determined that the behavior of the reverse traveling vehicle has not changed, the process proceeds to S450.

  In S450, the notification unit 50 outputs a command for performing an outward warning by the horn to the horn ECU 7. When this command is output, the horn ECU 7 outputs information indicating that the strong horn has been performed a predetermined number of times to the notification unit 50. And the alerting | reporting part 50 will complete | finish the alerting | reporting process regarding an outward warning, if this information is output from the alerting | reporting part 50. FIG.

[1-3. effect]
According to the first embodiment described in detail above, the following effects can be obtained.
(1a) Among a plurality of feature recognition processes that differ depending on the type of feature image, the features are sequentially one by one from the process of recognizing another feature image that has a smaller calculation amount than the image recognition that recognizes the vehicle front image as the feature image. Recognition processing is executed. In addition, when each feature recognition process is executed, a total likelihood value indicating the probability that the feature image can be recognized is obtained, and when it is determined that the total likelihood value exceeds the threshold value, the determination is made in this way. The forward vehicle thus determined is determined as a reverse vehicle. Further, when it is determined that the total likelihood value exceeds the threshold value, the execution of the feature recognition process with a larger calculation amount for the preceding vehicle is canceled. For this reason, it is possible to detect a reverse vehicle without necessarily performing feature recognition processing with a large calculation amount. Therefore, the reverse running vehicle can be detected more efficiently.

  (1b) One of the plurality of feature recognition processes is a process for recognizing a face image, which is an image indicating that the face of the occupant of the vehicle ahead is facing the host vehicle as a feature image. Here, since the process of recognizing a face image is relatively less affected by individual differences, the amount of calculation can be reduced compared to the process of recognizing a vehicle front image. For this reason, when the reverse running vehicle can be detected by the recognition of the face image, the execution of the process for recognizing the vehicle front image is canceled, so that the reverse running vehicle can be detected more efficiently.

  (1c) The likelihood calculation unit 30 sets a conversion value that becomes a large value according to the approach speed detected by the speed detection unit 25, and adds the set conversion value to the total likelihood value. Here, the possibility that the preceding vehicle is running backward increases as the approach speed increases. Further, since the approach speed calculation process is relatively simple, the calculation amount of the process can be greatly reduced as compared with the image recognition. For this reason, when the reverse vehicle can be detected by the approach speed calculation process, the execution of at least one image recognition process is cancelled, so that the reverse vehicle can be detected more efficiently.

  (1d) The determination unit 40 is at least one of the vehicle detection unit 20 and the likelihood calculation unit 30 when the road width of the target road is smaller than the reference width and the target road is not a one-way road. One process is stopped. On a road satisfying such a road condition, a forward vehicle facing the host vehicle is an oncoming vehicle that passes by the host vehicle, a parked vehicle, or the like, and is not a reverse vehicle. For this reason, it is possible to suppress an erroneous warning of a vehicle occupant or the like of a forward vehicle that is not a reverse running vehicle, and it is possible to reduce processing load by stopping unnecessary processing related to image recognition and the like.

  (1e) The alerting | reporting part 50 warns the passenger | crew of the reverse running vehicle determined by the determination part 40 using at least one of the inter-vehicle communication, the headlight of the own vehicle, and a horn. By performing an outward warning in this way, it is possible to increase the probability that the occupant of the reverse running vehicle will recognize that the vehicle is running backward.

[2. Other Embodiments]
As mentioned above, although the form for implementing this invention was demonstrated, this invention is not limited to the above-mentioned embodiment, It can implement in various deformation | transformation.

  (3a) In the above embodiment, the likelihood calculating unit 30 executes the feature recognition processing regarding these feature images in the order of the tail lamp non-existing image, the face image, and the vehicle front image. However, the present invention is not limited to this. is not. For example, the likelihood calculating unit 30 may execute the feature recognition process in the order of the face image, the tail lamp non-existing image, and the vehicle front image. That is, the likelihood calculation unit 30 may perform recognition of a face image as the first feature recognition and may recognize a tail lamp non-existing image as the second feature recognition.

  (3b) In the above embodiment, the likelihood calculating unit 30 calculates the converted value of the approach speed as the likelihood value P0 before the first feature recognition. However, the present invention is not limited to this. Absent. For example, the likelihood calculating unit 30 calculates the conversion value of the approach speed between the first feature recognition and the second feature recognition or between the second feature recognition and the third feature recognition. It may be added to the total value.

  (3c) In the above embodiment, when the determination unit 40 is a road whose road width is smaller than the reference width for the target road and the target road is not a one-way road, the vehicle detection unit 20 and the likelihood calculation Although at least one of the processes in the unit 30 is stopped, the present invention is not limited to this. For example, the determination unit 40 may stop at least one of the vehicle detection unit 20 and the likelihood calculation unit 30 when the host vehicle is traveling on a private road or a parking lot.

  (3d) A plurality of functions of one constituent element in the above embodiment may be realized by a plurality of constituent elements, or a single function of one constituent element may be realized by a plurality of constituent elements. . Further, a plurality of functions possessed by a plurality of constituent elements may be realized by one constituent element, or one function realized by a plurality of constituent elements may be realized by one constituent element. Moreover, you may abbreviate | omit a part of structure of the said embodiment. In addition, at least a part of the configuration of the above embodiment may be added to or replaced with the configuration of the other embodiment. In addition, all the aspects included in the technical idea specified only by the wording described in the claim are embodiment of this invention.

  (3e) In addition to the reverse running vehicle detection device 1 described above, a system including the reverse running vehicle detection device 1 as a component, a program for causing a computer to function as the reverse running vehicle detection device 1, and a semiconductor recording the program The present invention can also be realized in various forms such as a non-transition actual recording medium such as a memory and a reverse vehicle detection method.

 DESCRIPTION OF SYMBOLS 1 ... Reverse running vehicle detection apparatus, 2 ... Front camera, 3 ... Millimeter wave radar, 4 ... Navigation apparatus, 5 ... Inter-vehicle communication part, 6 ... Headlamp ECU, 7 ... Horn ECU, 8 ... Vehicle speed sensor, 10 ... Control , 11 ... CPU, 12 ... memory, 20 ... vehicle detection unit, 25 ... speed detection unit, 30 ... likelihood calculation unit, 40 ... determination unit, 50 ... notification unit

Claims (6)

If there is a travel lane that is the lane in which the host vehicle is traveling, or another lane that can be changed from the travel lane, the road that indicates the lane and the travel lane is set as the target road, and the target road A reverse running vehicle detection device for detecting a reverse running vehicle that is another vehicle running backward.
A vehicle detection unit configured to perform a process of detecting a forward vehicle that is another vehicle existing on the target road in front of the host vehicle;
A characteristic image characterizing each forward vehicle detected by the vehicle detection unit from an image obtained by an imaging device mounted on the host vehicle so as to capture the front of the host vehicle. A likelihood calculating unit configured to perform a process of calculating a likelihood indicating the likelihood that the feature image could be recognized by executing the feature recognition process,
It is determined whether or not the total likelihood value calculated by the likelihood calculating unit exceeds a predetermined threshold value, and the forward vehicle that has been determined that the total likelihood value exceeds the threshold value is the reverse traveling vehicle. A determination unit configured to perform a determination process as
A notification unit configured to perform processing for notifying at least a passenger of the own vehicle of the presence of the reverse running vehicle determined by the determination unit;
With
A plurality of the feature recognition processes are prepared according to the type of the feature image,
The likelihood calculating unit executes the plurality of feature recognition processes one by one in order from the feature recognition process with a small calculation amount for each of the preceding vehicles, and is calculated by executing each of the feature recognition processes. Configured to find the total likelihood,
The reverse traveling vehicle detection device configured to stop the processing of the likelihood calculating unit for the preceding vehicle that is determined that the total value of the likelihood exceeds the threshold value. The reverse vehicle detection device according to claim 1,
One of the plurality of feature recognition processes is a process of recognizing, as the feature image, a face image that is an image indicating that the face of an occupant of the preceding vehicle is facing the host vehicle. A reverse vehicle detection device configured. The reverse vehicle detection device according to claim 1 or 2,
A speed detection unit configured to detect an approach speed, which is a speed at which the preceding vehicle detected by the vehicle detection unit travels toward the host vehicle,
The likelihood calculation unit is configured to set a conversion value that becomes a large value according to the approach speed detected by the speed detection unit, and add the set conversion value to the total value of the likelihoods. Reverse running vehicle detection device. The reverse traveling vehicle detection device according to any one of claims 1 to 3,
The determination unit is a road whose road width is smaller than a predetermined reference width with respect to the target road, and when the target road is not a one-way road, the vehicle detection unit and the likelihood calculation unit A reverse vehicle detection device configured to stop at least one of the processes. The reverse traveling vehicle detection device according to any one of claims 1 to 4, wherein:
The informing unit is further configured to further perform a process of alerting the occupant of the reverse traveling vehicle determined by the determining unit using at least one of inter-vehicle communication and a headlight and a horn of the host vehicle. Traveling vehicle detection device. If there is a travel lane that is the lane in which the host vehicle is traveling, or another lane that can be changed from the travel lane, the road that indicates the lane and the travel lane is set as the target road, and the target road A reverse running vehicle detection method for detecting a reverse running vehicle that is another vehicle running backward.
A vehicle detection step for performing a process of detecting a forward vehicle that is another vehicle existing on the target road in front of the host vehicle;
For each forward vehicle detected by the vehicle detection step, a characteristic image characterizing the forward vehicle as the reverse traveling vehicle from an image obtained by an imaging device mounted on the own vehicle so as to image the front of the own vehicle A likelihood calculating step of performing a process of calculating a likelihood indicating the probability that the feature image could be recognized by executing the feature recognition process,
It is determined whether or not the total likelihood value calculated by the likelihood calculating step exceeds a predetermined threshold value, and the forward vehicle that has been determined that the total likelihood value exceeds the threshold value is the reverse traveling vehicle. A determination step for performing a determination process as
A notification step of performing a process of notifying at least the passenger of the host vehicle of the presence of the reverse running vehicle determined by the determination step;
With
A plurality of the feature recognition processes are prepared according to the type of the feature image,
In the likelihood calculating step, the plurality of feature recognition processes are executed one by one in order from the feature recognition process with a small calculation amount for each of the preceding vehicles, and the likelihood calculation step is calculated by executing each of the feature recognition processes. Find the total likelihood value,
The said determination process is a reverse running vehicle detection method which stops the process of the said likelihood calculation process about the said front vehicle judged that the total value of the said likelihood exceeds the said threshold value.

Images