JP5637735B2 - Pedestrian detection device - Google Patents

Description

  The present invention relates to a pedestrian detection device mounted on a vehicle, and more specifically to a pedestrian detection device that detects a pedestrian jumping out from a sidewalk onto a roadway.

  Usually, a pedestrian detection device mounted on a vehicle detects a pedestrian existing in front of the vehicle by searching for a preset reference pattern of the pedestrian from an image acquired by a camera.

  Here, the preset reference pattern of the pedestrian includes, for example, the direction of the pedestrian (forward, backward, leftward, rightward, etc.) and the size of the pedestrian (a pedestrian far away from the own vehicle, and the own vehicle). There are various patterns based on pedestrians in close proximity.

  Then, in the image acquired by the camera, a predetermined search window is arranged, the image in the search window is compared with the pedestrian reference pattern described above, and if the pedestrian reference pattern matches, the search is performed. It is determined that a pedestrian exists in the image in the window.

  As described above, the conventional pedestrian detection device detects a pedestrian from the image while moving a predetermined search window in the image acquired by the camera (see, for example, Patent Document 1).

JP 2008-152672 A

  However, in the conventional pedestrian detection device, a predetermined search window is arranged and moved in a wide area of the image acquired by the camera, and the image in the search window and preset various pedestrian reference patterns are arranged. Therefore, the amount of calculation for detecting a pedestrian from an image acquired by the camera is enormous.

  On the other hand, if the arrangement and movement of a predetermined search window are made rough in order to reduce the amount of calculation for detecting a pedestrian, the amount of calculation for detecting a pedestrian from an image acquired by the camera can be reduced. Unable to accurately detect pedestrians.

  Therefore, an object of the present invention is to provide a highly accurate and highly efficient pedestrian detection device that accurately detects a pedestrian from an image acquired by a camera and reduces the amount of calculation for detecting the pedestrian. It is to be.

In order to achieve the above object, a pedestrian detection device of the present invention is a pedestrian detection device mounted on a vehicle, an image acquisition means for acquiring an image around the vehicle, and an image acquired by the image acquisition means. A search density constant mode in which the search density is constant in the image area, a first area having a high search density, and a second area having a low search density. Based on the search density mode setting means for setting one of the search density mode and the search density emphasis mode for setting according to the road condition, and the search density mode set by the search density mode setting means The pedestrian search means for searching for a pedestrian from the image acquired by the above and the result notification means for notifying the result searched for by the pedestrian search means.
With this configuration, the search density for searching for a pedestrian from the image acquired by the image acquisition unit is constant in the search density constant mode in which the search density is constant in the image area, and the search density is high in the image area. In order to set one of the search density modes of the first area and the search density priority mode for setting the second area having a low search density according to the road conditions, the pedestrian can be accurately identified from the image acquired by the image acquisition means. It is possible to reduce the amount of calculation for detecting the pedestrian while detecting the pedestrian.

A preferable search density mode setting means sets one of the search density constant mode and the search density priority mode based on the possibility that a pedestrian is present on the roadway.
With such a configuration, a search density constant mode and a search density emphasis mode are set based on the possibility that a pedestrian is present on the roadway. A pedestrian can be detected with high accuracy.

A more preferable search density mode setting means sets the search density constant mode as the search density mode when a pedestrian is unlikely to exist in the road in the future, and the pedestrian may exist in the road in the future. If it is high, the search density emphasis mode is set as the search density mode.
With this configuration, particularly when there is a high possibility that a pedestrian will be present in the roadway in the future, the search density emphasis mode is set as the search density mode. Can be detected.

More preferably, the search density mode setting means sets one of the search density constant mode and the search density priority mode based on the state of the vehicle traffic light and / or the state of the pedestrian traffic light. Features.
With this configuration, either the search density constant mode or the search density priority mode is set based on the behavior of the vehicle and the pedestrian with respect to the state of the vehicle traffic light and / or the state of the pedestrian traffic light. Therefore, an optimal search density mode is set, and pedestrians can be detected efficiently, and pedestrians can be detected with high accuracy in a specific range that requires special attention.

More preferable search density mode setting means sets the search density constant mode as the search density mode when the state of the vehicle traffic signal is blue or red, and sets the search density priority mode when the state of the vehicle traffic signal is yellow. The search density mode is set.
With such a configuration, an optimal search is performed in order to set one of the search density constant mode and the search density priority mode based on the behavior tendency of the automobile and the pedestrian with respect to a specific state of the vehicle traffic signal. A density mode is set, and a pedestrian can be detected efficiently, and a pedestrian can be detected with high accuracy in a specific range that requires special attention.

Alternatively, the preferred search density mode setting means sets the search density constant mode as the search density mode when the state of the pedestrian traffic light is blue or red, and searches when the state of the pedestrian traffic light is blue flashing. The density emphasis mode is set as a search density mode.
With this configuration, the optimum search is performed because the search density constant mode and the search density priority mode are set based on the behavior tendency of the car and the pedestrian with respect to a specific state of the pedestrian traffic light. A density mode is set, and a pedestrian can be detected efficiently, and a pedestrian can be detected with high accuracy in a specific range that requires special attention.

Alternatively, the preferred search density mode setting means sets the search density constant mode as the search density mode when the state of the vehicle traffic light is blue or red and when the state of the pedestrian traffic light is blue or red, When the state of the vehicle traffic signal is yellow, or when the state of the pedestrian traffic light is blinking blue, the search density priority mode is set as the search density mode.
With this configuration, either the search density constant mode or the search density priority mode is set based on the behavior of the vehicle and the pedestrian with respect to the specific state of the vehicle traffic light and the state of the pedestrian traffic light. Therefore, an optimal search density mode is set, and a pedestrian can be detected efficiently. A pedestrian can be detected with high accuracy in a specific range that requires special attention.

Further, preferably, in the search density emphasis mode, the first region is around the boundary between the roadway and the sidewalk.
With this configuration, since the first region is around the boundary between the roadway and the sidewalk, it is possible to detect the pedestrian with high accuracy in a specific range where pedestrians need to be particularly careful to jump from the sidewalk to the roadway.

Further, preferably, in the search density emphasis mode, the search density for a pedestrian jumping from a sidewalk to a roadway is high.
With this configuration, in order to increase the search density for pedestrians jumping from the sidewalk to the roadway, it is possible to focus on pedestrians jumping from the sidewalk to the roadway and detect pedestrians with high accuracy.

  In order to achieve the above object, a pedestrian detection method of the present invention is a pedestrian detection method executed by a pedestrian detection device mounted on a vehicle, an image acquisition step for acquiring an image around the vehicle, and an image As for the search density for searching for pedestrians from the image acquired in the acquisition step, the search density constant mode in which the search density is constant in the image area, and the first area having the high search density in the image area and the search The search density mode selection step for selecting, according to the road condition, the search density mode selection step and the search density mode selection step for selecting one of the search density priority modes for setting the second region having a low density is selected. The search density mode setting step for setting the search density mode based on the result of the search, and the image acquisition scan based on the search density mode set in the search density mode setting step. Tsu including a pedestrian searching step of searching for a pedestrian from the image acquired by the flop, and the result notification step of notifying the result of the search by pedestrians searching step.

  Moreover, in order to achieve the said objective, each process which each structure of the pedestrian detection apparatus of this invention mentioned above can be grasped | ascertained as a pedestrian detection method which gives a series of processing procedures. This method is provided in the form of a program for causing a computer to execute a series of processing procedures. This program may be installed in a computer in a form recorded on a computer-readable recording medium.

  As described above, according to the pedestrian detection device of the present invention, it is possible to accurately detect a pedestrian from an image acquired by a camera and reduce the amount of calculation for detecting the pedestrian.

The block diagram which shows functionally the pedestrian detection apparatus 100 which concerns on one Embodiment of this invention. The flowchart which shows the pedestrian detection method 200 which the pedestrian detection apparatus 100 which concerns on one Embodiment of this invention performs. The figure which shows the front image of the vehicle imaged with the camera mounted in the vehicle The figure which shows a mode that the signal apparatus for vehicles is detected from the front image of the vehicle shown in FIG. The figure which shows the state transition of the traffic signal for vehicles, and the behavior tendency of a car and a pedestrian The figure which shows the search density mode with respect to the state of the traffic signal for vehicles The figure which shows a mode that a search density constant mode is set as a search density mode The figure which shows a mode that search density priority mode is set as search density mode Flowchart showing details of search density priority mode setting step S250 The figure which shows a mode that it sets so that the pedestrian who searches right may be searched intensively around the boundary of a roadway and a left sidewalk The figure which shows a mode that it sets so that the pedestrian who searches for the left may be searched intensively around the boundary of a roadway and a right sidewalk The figure which shows a mode that the traffic signal for pedestrians is detected from the front image of the vehicle shown in FIG. The figure which shows the state transition of the traffic signal for pedestrians, and the behavior tendency of cars and pedestrians The figure which shows the search density mode with respect to the state of the pedestrian traffic light

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram functionally showing a pedestrian detection apparatus 100 according to an embodiment of the present invention. In FIG. 1, the pedestrian detection apparatus 100 includes an image acquisition unit 110, a search density mode setting unit 120, a recording unit 130, a pedestrian search unit 140, and a result notification unit 150. The search density mode setting means 120 includes a traffic light state determination means 121 and a search density mode selection means 122.

  The image acquisition unit 110 captures a front image of the vehicle with a camera mounted on the vehicle, and acquires the image. Note that left and right and rear images of the vehicle as well as the front of the vehicle may be acquired according to the installation position, the number of cameras installed, and the viewing angle of the camera mounted on the vehicle. Here, it is assumed that a visible light camera and a near-infrared camera are mounted on the vehicle. The visible light camera acquires an image for detecting a traffic light described later, and the near-infrared camera acquires an image for searching for a pedestrian described later.

  The search density mode setting unit 120 sets a search density mode for searching for a pedestrian from the image acquired by the image acquisition unit 110. More specifically, the traffic signal state determination unit 121 determines the state of the traffic signal from the image acquired by the image acquisition unit 110. The search density mode selection unit 122 selects the search density mode based on the traffic light state determined by the traffic light state determination unit 121 in consideration of the possibility of a pedestrian jumping from the sidewalk to the roadway. Details of the search density mode will be described later.

  The recording unit 130 is, for example, a memory or the like, in which a reference pattern of a plurality of pedestrians and detailed setting information of a search density mode corresponding to a traffic light state are recorded. The search density mode setting unit 120 refers to the recording unit 130 and sets an appropriate search density mode.

  The pedestrian search unit 140 searches for a pedestrian from the image acquired by the image acquisition unit 110 based on the search density mode set by the search density mode setting unit 120.

  The result notification means 150 notifies the result searched by the pedestrian search means 140. For example, when a pedestrian is detected from the image acquired by the image acquisition unit 110, the pedestrian information is displayed on the monitor, and when it is determined that the pedestrian jumps out from the sidewalk to the roadway, a warning sound is generated. The driver may be notified by ringing.

  Next, the flow of processing executed by the pedestrian detection device 100 according to an embodiment of the present invention will be described.

  FIG. 2 is a flowchart showing a pedestrian detection method 200 executed by the pedestrian detection apparatus 100 according to an embodiment of the present invention. 2, the pedestrian detection method 200 includes an image acquisition step S210, a traffic light state determination step S220, a search density mode selection step S230, a constant search density mode setting step S240, a search density emphasis mode setting step S250, A pedestrian search step S260 and a result notification step S270 are included.

  In the image acquisition step S210, the image acquisition unit 110 captures a front image of the vehicle and acquires the image. Here, an image for detecting a traffic signal described later and an image for searching for a pedestrian described later are acquired. FIG. 3 is a diagram illustrating a front image of the vehicle imaged by a camera mounted on the vehicle.

  In the traffic light state determination step S220, the traffic light state determination means 121 of the search density mode setting means 120 detects a traffic light from the front image of the vehicle acquired in the image acquisition step S210. FIG. 4 is a diagram illustrating a state in which a vehicle traffic signal is detected from the front image of the vehicle illustrated in FIG. 3. In FIG. 4, the traffic light state determination means 121 detects a vehicle traffic signal in the traveling lane of the host vehicle from the front image of the vehicle acquired in the image acquisition step S210.

  And the traffic signal state determination means 121 determines the state of the detected vehicle traffic signal. Specifically, the traffic light state determination unit 121 determines whether the detected state of the vehicle traffic signal is blue, yellow, or red.

  In search density mode selection step S230, the search density mode selection means 122 determines whether or not the state of the vehicle traffic light determined in the traffic light state determination step S220 is yellow. When the state of the vehicle traffic signal is not yellow, the search density mode selection unit 122 selects the constant search density mode as the search density mode, and proceeds to the process of the constant search density mode setting step S240 (No in the search density mode selection step S230). ). On the other hand, when the state of the traffic signal for the vehicle is yellow, the search density mode selection unit 122 selects the search density emphasis mode as the search density mode, and proceeds to the processing of the search density emphasis mode setting step S250 (search density mode selection step). Yes at S230).

  Here, the state of the vehicle traffic signal and the behavior tendency of the automobile and the pedestrian will be described. FIG. 5 is a diagram showing the state transition of the traffic signal for vehicles and the behavior tendency of automobiles and pedestrians. As shown in FIG. 5, the vehicle traffic signal normally changes in order of blue → yellow → red → blue →.

  When the state of the vehicle traffic light is blue, the car travels in the driving lane and the pedestrian does not enter the roadway from the sidewalk (the pedestrian is stopped on the sidewalk). When the vehicle traffic light is red, the car is stopped and the pedestrian has entered the roadway from the sidewalk (the pedestrian crosses the pedestrian crossing). In other words, when the state of the vehicle traffic signal is blue and red, one of the automobile and the pedestrian is running (crossing) and the other is completely stopped.

  When the state of the vehicular traffic signal transitions from blue to yellow, the automobile may travel faster than the vehicular traffic signal is in the blue state. On the other hand, pedestrians tend to jump out from the sidewalk to the roadway in anticipation of the state of the vehicle traffic light changing to red, despite the possibility that the car may travel in a hurry (pedestrians) Crosses the pedestrian crossing). That is, when the state of the vehicle traffic signal is yellow, the automobile needs to pay attention especially to a pedestrian who jumps out from the sidewalk to the roadway.

  Based on the state of the vehicle traffic signal and the behavior tendency of the car and the pedestrian, the search density mode selection means 122 selects one of the search density constant mode and the search density priority mode. .

  FIG. 6 is a diagram illustrating a search density mode with respect to the state of the vehicle traffic signal. As described above, when the state of the traffic signal for the vehicle is yellow, the automobile needs to pay attention especially to the pedestrian jumping out from the sidewalk to the roadway. Therefore, the pedestrian detection device 100 is provided around the boundary between the roadway and the sidewalk. , Search for pedestrians with emphasis. That is, the search density mode selection unit 122 sets the search density to a constant value for the search density for searching for a pedestrian from an image acquired to detect the pedestrian when the state of the vehicle traffic signal is blue or red. Select the search density constant mode. On the other hand, the search density mode selection unit 122 increases the vicinity of the boundary between the roadway and the sidewalk with respect to the search density for searching for a pedestrian from an image acquired to detect the pedestrian when the state of the traffic signal for the vehicle is yellow. Select search density emphasis mode as density.

  In the search density constant mode setting step S240, the search density mode setting means 120 sets the search density constant mode as the search density mode based on the result selected in the search density mode selection step S230. FIG. 7 is a diagram illustrating a state in which the constant search density mode is set as the search density mode. In FIG. 7, the search density mode setting unit 120 sets the search density for searching for a pedestrian from the image acquired in the image acquisition step S <b> 210 so as to search for a pedestrian with a constant search density in the image area. To do.

  When the state of the vehicle traffic light is blue, the automobile is traveling in the driving lane and the pedestrian is stopped on the sidewalk. Therefore, the vehicle is focused on a specific range in the image area acquired in the image acquisition step S210. Instead of searching for a pedestrian, the search is set to search for a pedestrian with a constant search density within the image area. When the state of the vehicle traffic light is red, the car is stopped and the pedestrian crosses the pedestrian crossing. Similarly, in a specific range within the image area acquired in the image acquisition step S210, focus on the same. Rather than searching for pedestrians, settings are made so as to search for pedestrians with a constant search density within the image area.

  In search density emphasis mode setting step S250, search density mode setting means 120 sets the search density emphasis mode as a search density mode based on the result selected in search density mode selection step S230. FIG. 8 is a diagram illustrating how the search density priority mode is set as the search density mode. In FIG. 8, the search density mode setting unit 120 uses a specific range around the boundary between the roadway and the sidewalk in the image area for the search density for searching for a pedestrian from the image acquired in the image acquisition step S <b> 210. The search density is set to be high so as to search for pedestrians. On the other hand, as the search density is increased in the specific range, the search density outside the specific range is set low.

  Further, a method for increasing the search density in a specific range around the boundary between the roadway and the sidewalk will be described in more detail. FIG. 9 is a flowchart showing details of the search density priority mode setting step S250. In FIG. 9, the search density emphasis mode setting step S250 includes a search density emphasis range setting step S251 and a pedestrian search pattern setting step S252.

  In search density emphasis range setting step S251, search density mode setting means 120 sets a specific range for increasing the search density of pedestrians in the image area acquired in image acquisition step S210. For example, in FIG. 8, a specific range is set around each of the left and right boundaries between the roadway and the sidewalk.

  In the pedestrian search pattern setting step S252, the search density mode setting means 120 sets what kind of pedestrian reference pattern is searched in the specific range set in the search density priority range setting step S251.

  Specifically, in the vicinity of the boundary between the roadway and the left sidewalk, a pedestrian jumping from the sidewalk to the roadway is usually facing right. In other words, a right-handed pedestrian may be searched for mainly around the boundary between the roadway and the left sidewalk. FIG. 10A is a diagram illustrating a state in which a setting is made so that a right-handed pedestrian is preferentially searched around the boundary between the roadway and the left sidewalk. In FIG. 10A, the search density mode setting means 120 sets a specific range for increasing the search density of pedestrians within the image area acquired in the image acquisition step S210 as the vicinity of the boundary between the roadway and the left sidewalk, In the specific range, a reference pattern for a right-handed pedestrian is set to search for a right-handed pedestrian.

  Similarly, in the vicinity of the boundary between the roadway and the right sidewalk, it is only necessary to search for a pedestrian facing left. FIG. 10B is a diagram illustrating a state in which a setting is made so that a left-handed pedestrian is preferentially searched around the boundary between the roadway and the right sidewalk. In FIG. 10B, the search density mode setting means 120 sets a specific range for increasing the search density of the pedestrian within the image area acquired in the image acquisition step S210 as the vicinity of the boundary between the roadway and the right sidewalk, In the specific range, a reference pattern for a left-facing pedestrian is set, and a search for a left-facing pedestrian is focused.

  As described above, the pedestrian who jumps out from the left sidewalk to the roadway and the pedestrian who jumps out from the right sidewalk to the roadway are set to be searched with priority.

  Note that, within the image area acquired in the image acquisition step S210, a specific range for increasing the search density of pedestrians, and what pedestrian reference pattern to search for in the specific range is recorded. 130 may be set in advance. In the recording means 130, for example, a range of 2 [m] from the boundary line between the roadway and the sidewalk to the roadway side and the sidewalk side is set as a specific range, and among the reference patterns of a plurality of pedestrians, The reference pattern of which pedestrian is used to search for a pedestrian is recorded. The search density mode setting unit 120 selects a search density mode based on the state of the vehicle traffic signal, and further sets a specific search density corresponding to the selected search mode while referring to the recording unit 130.

  In the pedestrian search step S260, the pedestrian search means 140 is based on the search density constant mode set in the search density constant mode setting step S240 or the search density priority mode set in the search density priority mode setting step S250. A pedestrian is searched from the image acquired by the image acquisition means 110. Specifically, the pedestrian search means 140 arranges and moves a search window having a set search density, that is, a set coarseness, and an image in the search window and the set pedestrian reference pattern. And compare. In other words, in the constant search density mode, the pedestrian search means 140 arranges and moves the search window with the same coarseness in the image area acquired by the image acquisition means 110, and the image in the search window, Compare with the set pedestrian reference pattern. In the search density emphasis mode, the pedestrian search means 140 arranges and moves the search window with fine coarseness in the set specific range within the image area acquired by the image acquisition means 110, and outside the specific range. Then, the search window is arranged and moved with rough coarseness, and the image in the search window is compared with the set reference pattern of the pedestrian.

  In result notification step S270, the result notification means 150 notifies the result searched in pedestrian search step S260. For example, the image is displayed on the monitor so as to emphasize the image recognized as a pedestrian in the pedestrian search step S260, and among the images recognized as the pedestrian in the pedestrian search step S260, the pedestrian jumps from the sidewalk to the roadway. May be notified to the driver by sounding a warning sound.

  As described above, according to the pedestrian detection device 100 according to the embodiment of the present invention, when the state of the traffic signal for the vehicle is blue and red, the image is set by setting the constant search density mode as the search density mode. A pedestrian is searched with a constant search density within the image area acquired by the acquisition means 110. When the state of the vehicle traffic signal is yellow, the search density emphasis mode is set as the search density mode, so that a specific range around the boundary between the roadway and the sidewalk in the image area acquired by the image acquisition unit 110 is set. The search density is set high, and the search density in other ranges is set low so as to search for pedestrians. Thereby, the amount of calculation for detecting the said pedestrian can be reduced, detecting a pedestrian correctly from the image acquired by the image acquisition means 110. FIG.

  More specifically, if the traffic signal for vehicles, where pedestrians are likely to jump from the sidewalk to the roadway, is yellow, increase the search density around the boundary between the roadway and the sidewalk, and search density in other ranges. By lowering the pedestrian, it is possible to focus on detecting pedestrians jumping from the sidewalk to the roadway without increasing the amount of computation for detecting pedestrians.

  In the present embodiment, the search density mode is selected based on the state of the vehicle traffic signal. However, the present invention is not limited to this, and the search density mode is selected based on the state of the pedestrian traffic signal. It doesn't matter. FIG. 11 is a diagram illustrating a state in which a pedestrian traffic light is detected from the front image of the vehicle illustrated in FIG. 3. In FIG. 11, the traffic light state determination means 121 detects a pedestrian traffic light on the pedestrian crossing in the traveling lane of the host vehicle from the front image of the vehicle acquired by the image acquisition means 110.

  And the traffic signal state determination means 121 determines the state of the detected traffic signal for pedestrians. Specifically, the traffic light state determination unit 121 determines whether the detected state of the pedestrian traffic light is red, blue, or blue blinking.

  FIG. 12 is a diagram illustrating state transitions of pedestrian traffic lights and behavioral trends of automobiles and pedestrians. As shown in FIG. 12, the pedestrian traffic light normally changes in the order of red → blue → blue flashing → red →.

  When the state of the pedestrian traffic light is red, the automobile travels in the driving lane, and the pedestrian does not enter the roadway from the sidewalk (the pedestrian is stopped on the sidewalk). When the pedestrian traffic light is blue, the car has stopped and the pedestrian has entered the roadway from the sidewalk (the pedestrian crosses the pedestrian crossing). In other words, when the state of the pedestrian traffic light is red and blue, one of the car and the pedestrian is running (crossing) and the other is completely stopped.

  If the state of the pedestrian traffic light changes from blue to flashing blue, the car is ready to start and may even give up. On the other hand, pedestrians tend to jump out from the sidewalk to the roadway despite the possibility that the car will give up and start (the pedestrian crosses the pedestrian crossing). In other words, when the state of the pedestrian traffic light is blinking blue, the automobile needs to pay attention especially to pedestrians jumping from the sidewalk to the roadway.

  Based on the state of the pedestrian traffic light and the behavior tendency of the car and the pedestrian, the search density mode selection unit 122 selects one of the search density constant mode and the search density priority mode. To do.

  FIG. 13 is a diagram illustrating a search density mode with respect to the state of the pedestrian traffic light. As described above, when the state of the pedestrian traffic light is blinking blue, the automobile needs to pay attention particularly to the pedestrian jumping out from the sidewalk to the roadway. Search for pedestrians around the boundary. That is, when the state of the traffic light for the pedestrian is red or blue, the search density mode selection unit 122 sets the search density to a constant value for the search density for searching for a pedestrian from an image acquired to detect the pedestrian. Select the constant search density mode. On the other hand, when the state of the traffic light for pedestrians is yellow, the search density mode selection means 122 searches the vicinity of the boundary between the roadway and the sidewalk with respect to the search density for searching for pedestrians from the image acquired to detect the pedestrians. Select the search density emphasis mode to be high density.

  As described above, even if the search density mode is selected based on the state of the pedestrian traffic light, it goes without saying that the same effect as that obtained when the search density mode is selected based on the state of the vehicle traffic light can be obtained. Yes.

  Further, the search density mode may be selected based on both the state of the vehicle traffic light and the state of the pedestrian traffic light. Specifically, for example, it is assumed that the state of the vehicle traffic light is red and the state of the pedestrian traffic light is blinking blue. In this case, priority is given to the state of the pedestrian traffic light, and the search density mode selection means 122 selects the search density priority mode. In addition, it is assumed that the vehicle traffic light is yellow and the pedestrian traffic light is red. In this case, priority is given to the state of the vehicle traffic signal, and the search density mode selection means 122 selects the search density priority mode.

  That is, when the state of the vehicle traffic signal is blue or red and when the state of the pedestrian traffic light is blue or red, the search density mode selection unit 122 selects the constant search density mode. On the other hand, when the state of the traffic signal for the vehicle is yellow or when the state of the traffic light for the pedestrian is blinking blue, the search density mode selection unit 122 selects the search density priority mode.

  In this embodiment, the front image of the vehicle is acquired by the camera and the state of the traffic light is recognized from the image. However, the present invention is not limited to this. For example, a safe driving support system (DSSS: Driving Safety) The status of the traffic light may be recognized by Support Systems. Specifically, by notifying the vehicle of the state (signal information) of a traffic light from an optical beacon installed on the road, the vehicle acquires a front image of the vehicle with a camera and recognizes the state of the traffic light from the image. It does not have to be.

  In the present embodiment, the search density mode is selected based on the state of the traffic light in consideration of the possibility that the pedestrian jumps from the sidewalk to the roadway. However, the present invention is not limited to this. For example, in a pedestrian crossing where no traffic light is installed, it can be estimated that a pedestrian is likely to jump out of the sidewalk onto the roadway. In this case, the pedestrian crossing where the traffic light is not installed is recognized from the image acquired by the image acquisition unit 110, and the search density mode selection unit 122 is located in the pedestrian crossing when the vehicle travels on the pedestrian crossing. A search density emphasis mode for increasing the search density around the boundary between the roadway and the sidewalk may be selected.

  Furthermore, it can be estimated that there is a high possibility that a pedestrian will jump out of the traveling lane of the vehicle at the intersection. Similarly, in this case, the intersection is recognized from the image acquired by the image acquisition unit 110, and the search density mode selection unit 122, when the own vehicle enters the intersection, borders with other roads at the intersection. You may select the search density emphasis mode which makes search density high about the periphery.

  In the present embodiment, the search target is a pedestrian, but the present invention is not limited to this. For example, if a bicycle reference pattern is recorded in the recording unit 130 in advance, it goes without saying that the same effect can be obtained even if a bicycle is searched.

  In the present embodiment, the search density constant mode shown in FIG. 7 and the search density emphasis mode shown in FIGS. 8, 10A, and 10B are exemplified as search densities. However, the present invention is not limited to these. Absent. A mode based on other functions and heuristics may be set.

  INDUSTRIAL APPLICABILITY The present invention is useful for a pedestrian detection device that accurately detects a pedestrian jumping from a sidewalk to a roadway while reducing the amount of calculation for detecting the pedestrian.

DESCRIPTION OF SYMBOLS 100 Pedestrian detection apparatus 110 Image acquisition means 120 Search density mode setting means 121 Traffic light state determination means 122 Search density mode selection means 130 Recording means 140 Pedestrian search means 150 Result notification means 200 Pedestrian detection method S210 Image acquisition step S220 Traffic light state Determination step S230 Search density mode selection step S240 Search density constant mode setting step S250 Search density emphasis mode setting step S251 Search density emphasis range setting step S252 Pedestrian search pattern setting step S260 Pedestrian search step S270 Result notification step

Claims (10)

A pedestrian detection device mounted on a vehicle,
Image acquisition means for acquiring an image around the vehicle;
For search density of searching a pedestrian from the image acquired by said image acquiring means, and the search density constant mode to constant search density in the image area, in the image area, the search density of the search density constant mode One of the search density modes, which is a search density priority mode for setting a first area having a higher search density than the first search area and a second area having a lower search density than the search density in the constant search density mode, is set as a road condition. Search density mode setting means to set according to,
Based on the search density mode set by the search density mode setting means, a pedestrian search means for searching for pedestrians from the image acquired by the image acquisition means,
A pedestrian detection device comprising: result notification means for notifying the result searched for by the pedestrian search means.   The search density mode setting means sets the search density mode of the search density constant mode and the search density emphasis mode based on the possibility that a pedestrian is present on the roadway, The pedestrian detection device according to claim 1. The search density mode setting means includes:
When there is a low possibility that a pedestrian will be present on the roadway in the future, the search density constant mode is set as the search density mode,
The pedestrian detection device according to claim 2, wherein when there is a high possibility that a pedestrian is present on the roadway in the future, the search density emphasis mode is set as a search density mode.   The search density mode setting means sets either the search density constant mode or the search density priority mode based on the state of the traffic signal for vehicles and / or the state of the traffic light for pedestrians. The pedestrian detection device according to claim 3, wherein: The search density mode setting means includes:
When the state of the vehicle traffic signal is blue or red, the constant search density mode is set as the search density mode,
The pedestrian detection device according to claim 4, wherein when the state of the vehicle traffic signal is yellow, the search density priority mode is set as a search density mode. The search density mode setting means includes:
When the state of the pedestrian traffic light is blue or red, the constant search density mode is set as the search density mode,
The pedestrian detection device according to claim 4, wherein when the state of the pedestrian traffic light is blinking blue, the search density priority mode is set as a search density mode. The search density mode setting means includes:
When the state of the vehicle traffic light is blue or red, and when the state of the pedestrian traffic light is blue or red, the constant search density mode is set as a search density mode,
The search density priority mode is set as a search density mode when the state of the vehicle traffic signal is yellow or when the state of the pedestrian traffic signal is blinking blue. Pedestrian detection device.   The pedestrian detection device according to any one of claims 1 to 7, wherein, in the search density emphasis mode, the first region is around a boundary between a roadway and a sidewalk.   The pedestrian detection device according to claim 8, wherein in the search density priority mode, a search density for a pedestrian jumping from a sidewalk to a roadway is high. A pedestrian detection method executed by a pedestrian detection device mounted on a vehicle,
An image acquisition step of acquiring an image around the vehicle;
For search density of searching a pedestrian from the image acquired by the image acquisition step, a search density constant mode to constant search density in the image area, in the image area, the search density of the search density constant mode or search density than the set one of the search density mode and high search density emphasis mode search density than the search density of the search density constant mode and the first region is set to a lower second region A search density mode selection step to select according to road conditions;
A search density mode setting step for setting the search density mode based on the result selected in the search density mode selection step;
Based on the search density mode set in the search density mode setting step, a pedestrian search step for searching for a pedestrian from the image acquired in the image acquisition step;
A pedestrian detection method including a result notification step of notifying a result searched in the pedestrian search step.

Images