WO2010058821A1 - 接近物体検知システム - Google Patents

接近物体検知システム Download PDF

Info

Publication number
WO2010058821A1
WO2010058821A1 PCT/JP2009/069649 JP2009069649W WO2010058821A1 WO 2010058821 A1 WO2010058821 A1 WO 2010058821A1 JP 2009069649 W JP2009069649 W JP 2009069649W WO 2010058821 A1 WO2010058821 A1 WO 2010058821A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
vehicle
moving
approaching object
images
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/JP2009/069649
Other languages
English (en)
French (fr)
Japanese (ja)
Inventor
吉孝 内田
利幸 青木
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Faurecia Clarion Electronics Co Ltd
Original Assignee
Clarion Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Clarion Co Ltd filed Critical Clarion Co Ltd
Priority to US13/129,869 priority Critical patent/US8712097B2/en
Priority to EP09827602.5A priority patent/EP2369552B1/en
Priority to CN2009801458033A priority patent/CN102257533B/zh
Publication of WO2010058821A1 publication Critical patent/WO2010058821A1/ja
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/215—Motion-based segmentation
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10016—Video; Image sequence
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20068—Projection on vertical or horizontal image axis
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20112—Image segmentation details
    • G06T2207/20132—Image cropping
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30248—Vehicle exterior or interior
    • G06T2207/30252—Vehicle exterior; Vicinity of vehicle
    • G06T2207/30261—Obstacle

Definitions

  • the present invention relates to an approaching object detection system, and more particularly to a system for detecting an approaching object based on two or more images preceding and following a time series taken by a camera.
  • the monitor displays a focus on an area where the driver or other occupant visually perceives It is done to make up with the image.
  • Patent Documents 1 and 2 As a method of detecting a moving object approaching the host vehicle, for example, an area in which the position is changed between the images that sequentially precede and follow in time by optical flow processing based on a plurality of images that There is a method of detecting a part of an image as a moving object (Patent Documents 1 and 2).
  • a velocity vector (optical flow) for each pixel constituting an image is generated between two or more images (frames) that sequentially and sequentially Although it is necessary to obtain it, the velocity vector is specified by searching in all directions centering on the target pixel, and it is necessary to sequentially apply all the pixels constituting the image as the target pixel.
  • Patent Document 3 can be sufficiently coped with an on-board microcomputer because the calculation load is light, but the detection accuracy of the above-mentioned vertical edge is difficult, and erroneous detection increases, and practicability is poor. .
  • the present invention has been made in view of the above circumstances, and an object of the present invention is to provide an approaching object detection system capable of detecting an approaching moving object with high accuracy while reducing the load of arithmetic processing.
  • the approaching object detection system performs an optical flow process on an image so as to detect only a velocity vector in a specific direction, thereby reducing the processing load on the image while having a velocity vector in the specific direction.
  • the approaching object detection system is an optical system based on a camera which is fixed at a predetermined position and takes an image, and a plurality of images which are obtained by taking an image by the camera and which go back and forth in time series.
  • First moving area detection means for detecting (extracting) an image portion moved in a specific direction in time series among the images by flow processing, and the first moving area detection means for each of the plurality of images
  • Second moving area detection means for performing processing based on dynamic programming on each of the image portions detected by the step d) to obtain variation in size of the image portions along a direction different from the specific direction;
  • An approaching object determination unit that determines whether or not the moving object corresponding to the image portion is an approaching object with respect to the camera according to the change.
  • the specific direction is ideally a single (orientation)
  • the narrow angle range may be divided into four One direction, five directions, etc. may be used.
  • this specific direction is set in a large number, the computational load of the optical flow process will increase, so it is not preferable to carry out excessive fragmentation.
  • the above-mentioned dynamic programming is a method for determining the degree of similarity in consideration of expansion and contraction, which is used in optimal path search of so-called graph theory, and is also called DP (Dynamic Programming) matching. That is, by applying this processing based on dynamic programming to two similar images, it is possible to obtain enlargement / reduction ratios before and after scaling of these two images.
  • the “first direction different from the specific direction” is detected by the above-described first movement area detection means by optical flow processing. It is preferable that the direction is substantially orthogonal to the “specific direction” of the velocity vector to be attempted.
  • one of the multiple directions may be a reference direction. It is possible to apply substantially orthogonal directions.
  • the first moving area detection means performs optical flow processing based on a plurality of images which are sequentially back and forth in time
  • the second moving area detection means detects (extracts) the image portion moved to a specific direction in series, and for each detected image part, based on dynamic programming along a direction different from the specific direction.
  • the variation in size of the image portion in the later image in time series with respect to the image portion in the previous image in time sequence is determined, and the approaching object is determined according to the variation of the size.
  • the determination means determines whether or not the mobile corresponding to the image portion is approaching.
  • the optical flow process by the first moving area detection means is a process for detecting only the velocity vector in a specific direction
  • the operation processing load is reduced. It is possible to accurately detect (extract) an image part having a velocity vector in a specific direction, and the processing based on the dynamic programming by the second moving area detection means is a light processing as arithmetic processing, and the first The approaching object determination means performs the moving object detection by subjecting the image portion detected by the movement area detection means to processing based on dynamic programming in a direction different from the specific direction described above to obtain variation in size. It is possible to accurately determine whether or not the vehicle is approaching.
  • the second moving area detection means is a dynamic programming method for each image portion detected by the first moving area detection means in each of the plurality of images.
  • the fluctuation of the magnitude is obtained based on the correspondence of the distribution of signal values.
  • the second moving area detecting means is for each image portion detected by the first moving area detecting means in each of the plurality of images.
  • a distribution (profile) of signal values along a direction different from the specific direction for each image portion by performing processing based on dynamic programming, and Distribution of signal values (distribution of signal values along a direction different from a specific direction) and distribution of signal values of an image portion in a later image in time series (signal values along a direction different from a specific direction)
  • the ratio of the sizes of the two image parts can be easily determined by correlating the distribution with the distribution), and the ratio of the sizes is considered as the “variation in size” in the approaching object detection system according to the present invention. Because you can .
  • the first moving area detection means is a direction corresponding to the horizontal direction in the image as the specific direction to be subjected to the optical flow processing, and the horizontal direction. And one or more predetermined orientations within an angle range from 15 degrees upward to 15 degrees downward, and the second moving area detection means is based on the dynamic programming method.
  • a direction corresponding to the vertical direction in the image is applied as a direction different from the specific direction in processing.
  • the first moving area detection unit is an orientation corresponding to the horizontal direction in the image as the specific orientation to be subjected to optical flow processing. Since the orientation within the angle range of 15 degrees upward to 15 degrees downward with respect to the horizontal direction is applied, it is possible to detect a mobile body moving mainly in the horizontal direction.
  • the processing of the dynamic programming by the second moving area detecting means is an image of the image portion detected by the first moving area detecting means, with respect to the image portion including the moving body moving mainly in the horizontal direction.
  • the approaching object determination means determines that the moving object in the detected image portion is a moving object (approaching object) approaching the camera.
  • the approaching object determination means determines that the moving object in the detected image portion is a moving object moving away from the camera.
  • the approaching object determination means determines that the moving object in the detected image portion is approaching with respect to the diameter direction of the lens of the camera but is separating with respect to the optical axis direction. It is determined that the object is not an approaching object.
  • the approaching object determination means determines that the moving object in the image portion is an approaching object when the magnitude of the change is in the increasing direction.
  • the approaching object determination unit detects the detection result of the second moving area detection unit only when it fluctuates in the direction in which the size increases. Since the moving object in the image portion is determined to be an approaching object, the accuracy of determining the approaching object can be improved more than that of determining the approaching object based only on the detection result by the first moving area detection means. .
  • the first moving area detection means may, among the images obtained by photographing by the camera prior to detection of the moved image portion by the optical flow processing, It is preferable to exclude at least one of the sky and the ground.
  • the sky and the ground are excluded from the target of the approaching object to be detected by the present invention.
  • the calculation load of the optical flow process can be further reduced.
  • the first moving area detection means Since the sky and the ground in the image are usually areas where the density (brightness) and hue are substantially uniform, the first moving area detection means has such a flat (brightness) distribution (brightness) A region (generally uniform) is judged as an empty region or a region of the ground, and such a concentration (brightness) distribution is substantially flat (a density (brightness) is substantially uniform) prior to optical flow processing.
  • the amount of computation can be reduced by removing the signal from the optical flow processing.
  • the first moving area detection means performs optical flow processing based on the two images (two frames) preceding and following in time series, and time series of the images
  • the second moving area detection means detects an image portion moved in a specific direction, and processing based on dynamic programming based on the two images (two frames) targeted for the optical flow processing. It is preferable that the variation of the magnitude is determined by
  • the approaching object detection system having the preferable configuration as described above, the approaching object can be detected with high accuracy based on only two images which are sequentially back and forth.
  • the camera is fixed to the rear of the vehicle so as to capture an image of the rear of the vehicle.
  • the approaching object detection system for example, when the vehicle is parked by moving forward instead of moving backward, the vehicle is withdrawn from the parking lot or the garage when moving out.
  • the camera in the approaching object detection system according to the present invention is fixed to the rear of the vehicle, it is difficult to visually recognize this. Since the system reliably detects an approaching object approaching a moving vehicle, the driving operation of the vehicle can be safely supported.
  • the camera is attached to a vehicle, and vehicle information acquisition means for acquiring vehicle information on the traveling state of the vehicle including the vehicle speed, and the vehicle information indicates movement of the vehicle Image processing that performs size correction to correct the size of the image in a direction that suppresses the occurrence of a change in the size of the image according to the movement of the vehicle in the images that sequentially move in time series. It is preferable to have a part.
  • the image processing unit when the vehicle equipped with the camera moves, the image processing unit performs size correction to move the vehicle, thereby moving on the image.
  • the image processing unit performs size correction to move the vehicle, thereby moving on the image.
  • the image processing unit changes the enlargement ratio of the image according to the moving distance of the vehicle between the images preceding and following in time series at the time of size correction.
  • the enlargement ratio is changed according to the vehicle movement distance between the images one after another in time series.
  • the first moving area detecting means obtains a variation in shape of an area occupied by the moving object moved in the specific direction between the images sequentially back and forth in time series
  • the approaching object determination unit determines that the moving body is a pedestrian when the variation in the shape of the area is larger than a preset setting value.
  • the approaching object detection system having such a preferable configuration, it is possible to determine the pedestrian based on the variation in the shape of the area occupied by the moving object on the image. That is, in the case of a pedestrian, the moving speed is slow or the moving direction is unstable, and the appearance of the optical flow in a specific direction is not remarkable between the images that are back and forth in time series as compared with the vehicle. In some cases, the size variation due to dynamic programming may not be noticeable. However, pedestrians shake their arms or step on their feet, and the moving direction of a part of the body changes, which causes variations in the coordinates obtained by the optical flow in a specific direction, whereas In the case of such a constant shape, the variation in shape is small.
  • the directivity of the optical flow is specified by determining the moving object as a pedestrian, and the processing load using the dynamic programming method is light. Even with the means, pedestrians can be detected with high accuracy.
  • the variation calculating unit obtains a difference in the area of the area occupied by the moving object moved in the specific direction in the images sequentially back and forth, and the plurality of images back and forth in time series It is preferable to set the value accumulated over the range as the variation.
  • a pedestrian can be obtained by accumulating the difference in area of the area occupied by the moving object between the images which are sequentially back and forth in a plurality of images. Even if the moving speed is low, and the shape change is small between the images adjacent in time series, the pedestrian can be determined with high accuracy.
  • the approaching object detection system it is possible to accurately detect an approaching moving object while reducing the load of arithmetic processing.
  • FIG. 1 is a block diagram showing a configuration of an approaching object detection system 100 according to Embodiment 1 of the present invention. It is a schematic diagram which shows the typical example of the condition where the approaching object detection system 100 shown in FIG. 1 is utilized. 5 is a flowchart showing the processing procedure of the approaching object detection system 100. It is a figure showing the example of the time-sequential image image
  • the obtained optical flow and an area moving in a substantially horizontal direction obtained based on the optical flow are shown.
  • This is a schematic diagram showing a moving state determined to be an approaching object, a moving state determined to be an distant object, and a moving state determined to be neither approaching nor separating from the vehicle. is there.
  • FIG. 17 is a flowchart illustrating the first half of the processing procedure of the approaching object detection system 100b according to the second embodiment.
  • FIG. 17 is a flowchart showing the second half of the processing procedure of the approaching object detection system 100b according to the second embodiment.
  • FIG. 1 is a block diagram showing a configuration of an approaching object detection system 100 according to Embodiment 1 (Example 1) of the present invention
  • FIG. 2 is a typical situation where the approaching object detection system 100 shown in FIG.
  • FIG. 3 is a schematic diagram showing a typical example, showing a situation where a vehicle 200 equipped with this approaching object detection system 100 is about to enter an intersection
  • FIG. 3 is a flowchart showing the processing procedure of the approaching object detection system 100.
  • the vehicle 200 on which the approaching object inspection system 100 is mounted is hereinafter referred to as a host vehicle 200 in order to distinguish it from other vehicles 300 described later.
  • the illustrated approaching object detection system 100 is a camera 10 fixed to the front end of the vehicle 200 so as to capture a landscape Jm on the side (right side) of the vehicle at the front end of the vehicle 200 as an image P (t). And an image P (t) obtained by optical flow processing based on two images P (t) and P (t + 1) (refer to FIG. 4) which are obtained by photographing with this camera 10 and which are sequentially obtained in time series. , P (t + 1), rectangular image portions Q (t), Q (t) including regions q (t), q (t + 1) (see FIG. 6) moved in a substantially horizontal direction (specific direction) in time series.
  • first moving area detecting means 30 for detecting (extracting) t + 1) see FIG.
  • the approaching object detection system 100 includes the own vehicle 200 at the front end of the own vehicle 200 in addition to the camera 10 that captures the landscape Jm on the right side of the own vehicle 200.
  • the camera 20 which photographs the scenery Jn of the left side of is also provided.
  • a single camera having an optical system with a wide angle of view of, for example, 180 degrees or more may be applied, and in that case, the right side of the captured image
  • An image obtained by cutting out the angle of view range corresponding to the other landscape Jm may be applied as an image obtained by shooting with the camera 10 for the right side, and among the photographed images, the landscape in the left direction
  • An image obtained by extracting the angle of view corresponding to Jn may be applied as an image obtained by photographing with the left-side camera 20.
  • the operation of the approaching object detection system 100 will be described.
  • the driver of the vehicle interior of the vehicle 200 advances the vehicle 200.
  • Other vehicles, bicycles, etc. passing through the road crossing the road may be overlooked or difficult to see directly.
  • the camera 10 installed at the front end can capture a landscape Jm on the right side of the intersecting road, and The camera 20 installed at the front end captures a landscape Jn on the left side of the intersecting road (step # 1 in FIG. 3).
  • the scenery Jm on the right side taken by the camera 10 is performed, and the scenery Jn on the left side taken by the camera 20 is the scenery Jm on the right side.
  • the description is appropriately omitted because it may be applied symmetrically.
  • the image captured by the camera 10 is acquired as an image P of 30 [frames / sec] in the image acquisition unit 25 with, for example, VGA size (640 [pix] x 480 [pix]) and video signal standard NTSC. (# 2 in FIG. 3).
  • the images P (t), P (t + 1), P (t + 2),... Which are photographed by the camera 10 and acquired sequentially by the image acquisition unit 25 are input to the first moving area detection means 30.
  • the first moving area detection means 30 performs optical flow processing on the basis of the images P (t), P (t + 1), P (t + 2),.
  • the first moving area detection unit 30 detects an interval between two images P (t) and P (t + 1) that are consecutive in time series (hereinafter referred to as “image P (t + 1), P (t + 2) , And between images P (t + 2) and P (t + 3),...
  • the direction for obtaining the optical flow is limited to the substantially horizontal direction. That is, when a moving body is usually extracted by optical flow processing, as shown in FIG. 5A, the optical flow is obtained for the direction of 360 degrees centering on the pixel of interest as the direction for obtaining the optical flow.
  • an angle interval is, for example, an interval of 1 degree or an interval of 5 degrees, and in the case of an interval of 1 degree, it is necessary to perform 360 times of search per pixel of interest. It is necessary to perform the search 72 times per one target pixel ((a) in the figure shows the directions at intervals of 10 degrees).
  • the first moving area detection means 30 in the present embodiment horizontally (in the directions of 0 degree and 180 degrees) and the horizontal direction with respect to the pixel of interest.
  • Optical flow is to be obtained for only a total of six directions of an upward direction of 15 degrees with respect to the direction and an downward direction of 15 degrees with respect to the horizontal direction.
  • the first movement area detection unit 30 in the embodiment performs only six searches per pixel of interest.
  • the computational load required for optical flow processing can be significantly reduced compared to the conventional computational load for performing optical flow processing for an angle of 360 degrees around the entire circumference.
  • the calculation microcomputer used for the first moving area detection means 30 can be covered by a general on-vehicle microcomputer, and a reduction in manufacturing cost can also be realized.
  • an object moving in a substantially horizontal direction in the image P is It can be detected (extracted) as a mobile. That is, since other vehicles 300 and road traffic objects such as bicycles and pedestrians are moving objects that move generally in the horizontal direction, the other vehicles 300 that are passing as the above-described moving objects in the substantially horizontal direction And bicycles, pedestrians, etc. can be detected without leakage.
  • direction to be subjected to the optical flow processing is limited to six directions in the present embodiment, 14 directions having a slightly finer 5 degree interval, and other angle intervals (equal angle intervals) May be applied).
  • the moving object to be detected moves in a substantially horizontal direction, such as another vehicle 300, a bicycle, a pedestrian, etc.
  • a substantially horizontal direction such as another vehicle 300, a bicycle, a pedestrian, etc.
  • the moving object to be detected is an object that moves from the upper side to the lower side or from the lower side to the upper side (falling objects, raindrops, hanging load of crane, etc.)
  • the direction of movement in the substantially vertical direction may be limited to the substantially vertical direction as the target direction of the optical flow processing.
  • the optical flow OP (t to t + 1) and the images P (t + 1) and P (t + 2) of FIG.
  • the optical flow OP (t + 1 to t + 2) of (b) is obtained.
  • the gray colored portions on the left side are regions having velocity vectors (optical flows) in the substantially horizontal direction.
  • the first moving area detection means 30 occupies the moving object in the substantially horizontal direction in each of the images P (t) and P (t + 1). Regions q (t) and q (t + 1) are identified (# 3 in FIG. 3).
  • the first moving area detection means 30 moves the mobile unit in the substantially horizontal direction in each of the images P (t + 1) and P (t + 2).
  • the occupied regions q (t + 1) and q (t + 2) are identified (# 3 in FIG. 3).
  • the colored portions on the right side are the regions q occupied by the mobile body in the substantially horizontal direction.
  • the first moving area detection unit 30 encloses the area q (t) in the image P (t) in a rectangle and sets it as an image portion Q (t) including the moving area, and similarly
  • the area q (t + 1) in the image P (t + 1) is enclosed in a rectangle, which is set as an image portion Q (t + 1) including the movement area (see FIG. 7A).
  • the area q (t + 2) in) is enclosed in a rectangle, and this is set as an image portion Q (t + 2) including the movement area (see FIG. 7A, # 4 in FIG. 3).
  • the image portions Q (t), Q (t + 1), Q (t + 2),... Detected by the first moving area detecting means 30 are inputted to the second moving area detecting means 40.
  • the timing when the image portion Q is input from the first moving area detection unit 30 to the second moving area detection unit 40 is sequentially input as soon as the first moving area detection unit 30 identifies the image portion Q. Be done.
  • the second moving area detection means 40 detects image portions Q sequentially input and output between two image portions Q (t) and Q (t + 1), image portions Q (t + 1), DP matching processing is performed between Q (t + 2), ..., ... (# 5, # 6, # 7 in Fig. 3).
  • an image signal value (brightness along the vertical direction at a predetermined position preset in the horizontal direction)
  • the distribution (profile) of pixel values (values etc.) is determined (# 5 in FIG. 3), and then, for each image portion in a set of two images sequentially linked in time series, relative as shown in FIG.
  • the distribution of the image signal values of the image portion Q (t + 2) determined in step S2 is matched (# 6 in FIG. 3), and matching based on this association results in the image portion Q (t + 2) of both image portions Q (t + 1)
  • processing based on DP matching is sequentially performed on each image portion of another set of two images sequentially linked in time series, and the obtained enlargement factor K is sequentially input to the approaching object determination unit 50.
  • the approaching object determination means 50 determines an area q included in the image portion Q (an image of the moving body (the other vehicle 300 is illustrated in the figure)) based on the enlargement factor K detected by the second moving area detection means. Is a moving object approaching or moving away from the host vehicle 200, or approaching in a substantially horizontal direction, but along the optical axis direction of the camera 10. It is determined that the moving object is moving away (for example, a co-traveling vehicle traveling in parallel) and not a moving object approaching as a whole (# 8, # 9, # 10, # 11 in FIG. 3). , # 12).
  • the time series is relatively later among the two images P (t) and P (t + 1) that are sequentially related in time series.
  • the size of the image portion Q (t + 1) in the image P (t + 1) is larger than that of the image portion Q (t) in the previous image P (t) because the time series is relatively large.
  • Can be determined to have an increasing tendency (# 8 in FIG. 3) and the moving object corresponding to the region q included in the image portion Q approaches the vehicle 200 as shown in A1 of FIG. It is determined that the object is moving (for example, an approaching vehicle) (# 9 in FIG. 3).
  • the image P (t + 1) in the later time series is Since the size of the image portion Q (t + 1) is smaller than that of the image portion Q (t) in the previous image P (t) in time series, it is determined that the size does not tend to be equal. 3 and the moving object corresponding to the area q included in the image portion Q is determined to be an object moving away from the host vehicle 200, as shown at A3 in FIG. (# 12 in FIG. 3).
  • the first moving area detection unit 30 performs the optical flow processing based on the two images P and P sequentially related in time series.
  • the second moving area detection means 40 detects (extracts) the image portions Q, Q which have been moved in a substantially horizontal direction in time series among the images P, P, and the second moving area detection means 40 By performing processing based on dynamic programming along the vertical direction, the image portion Q (t) in the previous image P (t) in time-sequentially the image P (t + 1) in time-sequentially The variation in size of the image portion Q (t + 1) is obtained, and the approaching object determination means 50 determines the moving object corresponding to the image portion Q (t), Q (t + 1) according to the variation in the size. It is determined whether it is approaching.
  • the optical flow processing by the first moving area detection means 30 is processing for detecting only the velocity vector in the substantially horizontal direction, it is possible to substantially reduce the processing load.
  • the image portion Q corresponding to another vehicle 300 moving in the horizontal direction can be detected (extracted) with high accuracy, and the processing based on the dynamic programming by the second moving area detection means 40 is a processing with a light computational load Further, by performing processing based on dynamic programming in the vertical direction on the image portion Q detected by the first moving area detection means 30, the approaching object determination means 50 is obtained by obtaining a variation in size. However, it can be accurately determined whether the moving object is approaching.
  • each of the second moving area detection means 40 is an image detected by the first moving area detection means 30 in each of the two images P and P. Processing based on dynamic programming is performed on parts Q and Q, and for each image part Q and Q, the distribution (profile) of signal values along the vertical direction is obtained, and the previous image is obtained in chronological order The distribution of the signal values of the image portion Q (t) in P (t) and the distribution of the signal values of the image portion Q (t + 1) in the image P (t + 1) later in time series correspond to each other. By doing this, the size ratio between the two image parts Q (t) and Q (t + 1) can be easily determined, which makes it easy to determine whether the moving object is a truly approaching object or not. It can be determined.
  • the approaching object determination unit 50 determines the inside of the passenger compartment.
  • a signal S is output to warn or warn a driver or other occupant of the vehicle.
  • the image display device and the notification device in the existing in-vehicle system (for example, a car navigation system, a voice guidance system, a camera monitor system, a back alarm sound alerting system, etc.) having received this signal S It may be displayed or sound may be generated to alert or warn.
  • the existing in-vehicle system for example, a car navigation system, a voice guidance system, a camera monitor system, a back alarm sound alerting system, etc.
  • a rectangular frame T surrounding an area q representing an approaching object blinks or the frame T is visually stimulated. It is also possible to employ display in a vivid color or the like.
  • the approaching object detection system 100 has been described only for the landscape Jm taken on the right side by the camera 10, the landscape Jm taken on the left side taken by the camera 20 is the scenery Jm on the right side And should be applied symmetrically.
  • the first moving area detection means 30 is obtained by photographing with the camera 10 prior to detection of the moved area q (image part) by optical flow processing.
  • the image P at least one of the region corresponding to the sky and the region corresponding to the ground may be excluded in advance.
  • these empty regions are extracted from the image P before the optical flow processing. By removing the ground region in advance, it is possible to further reduce the calculation load of the optical flow processing.
  • the first moving area detection means Since the sky and the ground in the image are usually areas where the density (brightness) and hue are substantially uniform, the first moving area detection means has such a flat (brightness) distribution (brightness)
  • the optical flow is determined by determining that the density (brightness) is substantially uniform area as an empty area (represented by a symbol Ps in FIG. 4A) or a ground region (represented by a symbol Pr in FIG. 4A). The amount of calculation can be reduced by removing such a region where the concentration (brightness) distribution is substantially flat prior to processing so as not to be a target of optical flow processing.
  • the first moving area detection unit 30 performs time-sequentially of the images P by optical flow processing based on two images (two frames) that sequentially and sequentially.
  • the second moving area detecting means 40 detects the area q moved to the specific direction in the second direction, and the second moving area detecting means 40 performs processing based on the dynamic programming based on the two images (two frames) targeted for the optical flow processing.
  • the approaching object detection system 100b according to the second embodiment is a modification of the first embodiment, and the same components as those in the first embodiment are denoted by the same reference numerals as those in the first embodiment, and the description thereof is omitted.
  • the differences from Embodiment 1 will be mainly described.
  • the approaching object detection system 100b includes a camera 240, an image processing unit 210, and a variation calculation unit 220 in addition to the configuration shown in the first embodiment.
  • the camera 240 is fixed to the rear of the host vehicle 200 and captures an image of the rear of the host vehicle 200.
  • the image processing unit 210 corrects the image P obtained by the image acquisition unit 25 based on the vehicle information obtained from the vehicle information acquisition unit 230 and outputs the image P to the first moving area detection unit 30 b.
  • the vehicle information acquisition unit 230 acquires vehicle information via CAN communication or the like from a control unit (not shown) that controls the traveling of the vehicle.
  • the vehicle information includes shift information indicating a shift position of the vehicle, vehicle speed information for obtaining a movement distance of the vehicle, and steering angle information for obtaining a rotational component of the vehicle.
  • the vehicle speed information may be obtained not only from the signal from the vehicle speed sensor outside the figure but also from the signal from the wheel speed sensor outside the figure at low speed when the detection accuracy of the vehicle speed sensor is low.
  • FIG. 12 is an explanatory view showing a relationship between an actual image photographed by the cameras 10 and 20 and an image P projected on a two-dimensional plane, and as shown in this figure, the cameras 10 and 20 at the position of 0c.
  • the real image (three-dimensional camera coordinates Pc) captured by the above is converted into an image P on the screen projected on a two-dimensional plane.
  • the object image projected on the image P is enlarged.
  • FIG. 13 A specific example will be described with reference to FIG. 13.
  • the object image B (a) in the image P (a) shown in FIG. 13 (a) is obtained before the host vehicle 200 moves forward (this object image It is assumed that the vehicle indicated by B (a) is at a stop), and after the own vehicle 200 approaches the camera coordinates Pc, an object image B (b) on the image P (b) as shown in FIG. b) is enlarged.
  • the image processing unit 210 based on the vehicle speed information obtained by the vehicle information obtaining unit 230, for example, an image P (a) shown in FIG. 13 (a) and an image P (shown in FIG. Calculate the movement amount of the vehicle 200 between b) and use it as a distance correction value of perspective projection (coordinate conversion), and the projection size of the object image B (b) in the current image P (b), A correction is performed to obtain an image P (c) in which the enlargement ratio of the image P is reduced so that the projection size of the object image B (a) of the image P (a) in the previous frame is the same.
  • the sizes of non-moving objects such as the background in each image P can be made uniform, and only moving objects are enlarged, and the approaching object image is enlarged. Distant object images can be reduced.
  • the lens distortion correction is a correction that eliminates or reduces distortion caused by the lens characteristics of the cameras 10, 20, and 240. That is, depending on the lens, images of the lens peripheral portion and the lens focusing portion may be different.
  • FIG. 14 (b) shows an example in which barrel distortion type distortion appears, and in this case, distortion occurs in which the frame wx, wy is curved. Therefore, when the lens has such distortion, the frame frames wx and wy are straight as shown by dotted lines in the figure, respectively.
  • Image processing (lens distortion correction) to form
  • the relationship between the size of the object on the image P and the distance to the vehicle 200 can be made uniform at any position on the image P.
  • the first moving area detection unit 30 b performs optical flow processing on the time-series continuous image P after the correction in the image processing unit 210, and moves the image in the specific direction in time series. Detect part Q. This optical flow process is the same as that of the first embodiment, and the description will be omitted.
  • the variation calculating section 220 obtains the variation in the shape of the area q occupied by the moving body.
  • the variation in shape is used for pedestrian determination in the approaching object determination means 50b described later.
  • the difference in the area of the area q occupied by the moving object between the images P that precede and follow in time series is determined, and this difference is set in advance Use the value accumulated over the interval. That is, when the approaching object is a vehicle, the shape is constant, and the change in the area between the preceding and succeeding frames is also small. On the other hand, in the case of a pedestrian, the movement direction of the hand or foot changes, so the size of the area q having the optical flow in the horizontal direction changes, but in the case of a pedestrian Therefore, the change between the previous and next frames is small.
  • the value becomes larger as the area q in which the shape (area) changes continuously between the frames, and this value is the degree of the variation in the shape of the area q. It can be a value representing
  • the approaching object determination unit 50b performs pedestrian determination in addition to the determination of the approaching vehicle, the parallel traveling vehicle, and the distant vehicle described in the first embodiment.
  • the approaching object determination unit 50b extracts, as the image portion Q, a pedestrian or another vehicle 300 having a high possibility size as a determination target object, and extracts this determination target object. If the image portion Q has a variation in shape larger than a preset value (degree), it is determined as a pedestrian. In this case, it is determined whether or not the accumulated value of the area difference between the regions q in the preceding and succeeding images P is larger than a preset threshold value.
  • the enlargement ratio is calculated based on the enlargement ratio K obtained by the second moving area detection unit 40.
  • K is greater than 1
  • the enlargement ratio K is an equal ratio of approximately 1, parallel traveling vehicles traveling parallel to the host vehicle 200 If the magnification ratio K is less than 1, it is determined that the vehicle is a distant vehicle (moving object).
  • the approaching object detection process is performed at the time of going into an intersection at the time of forward traveling or at the time of backward traveling as shown in FIG. 2 described above. Or may be automatically started when a backward travel is detected, or may be started by the driver turning on a switch instructing to start an approaching object detection process.
  • intersection approach at the time of forward traveling may be regarded as intersection approach when the vehicle speed falls below the set speed by the driver's braking operation to simplify control, or navigation outside the figure
  • intersection approach may be determined based on information of the system and other driving support systems.
  • step S1 the vehicle information acquisition unit 230 acquires vehicle information, and the process proceeds to the next step S2.
  • the image P is acquired 30 frames per second.
  • step S4 after the size correction and lens distortion correction described above are performed in the image processing unit 210, the process proceeds to step S5.
  • step S6 the variation calculation unit 220 of the first moving area detection unit 30b determines the variation in shape of the area q occupied by the moving object obtained in step S5 (accumulation of the area difference between the preceding and succeeding images). Ask.
  • the approaching object determination means 50b determines the presence or absence of a pedestrian and the approaching vehicle, the parallel traveling vehicle, and the distant vehicle.
  • step S11 it is determined whether or not the image part Q is large to some extent. If it is large, the process proceeds to step S12. If it is not large, the process proceeds to step S14.
  • “somewhat large” means that the size of the image portion Q on the image P is high enough to avoid the collision with the vehicle 200, that is, the image portion on the image P
  • the size of Q corresponds to the size of the moving object itself and the distance to the vehicle 200. For example, when the moving object is another vehicle 300, although it has a certain size itself, it has a high moving speed, so it is from a time point where it is present to a certain distance (small in image P). Care must be taken to avoid contact.
  • the size of the moving object is smaller than that of the other vehicles 300, but is relatively close to the vehicle 200 because the moving speed is low (the size on the image P is large) At this point, care must be taken to avoid contact.
  • a threshold value of the size on the image P that needs to be recognized as an approaching object is set in advance, regardless of whether the moving object is another vehicle 300 or a pedestrian. In step S11, the threshold and the size of the area q are compared and determined.
  • step S12 the variation in the shape of the area q occupied by the moving object obtained in step S6 (the accumulated value between the set frames of the area difference in the previous and subsequent images P in the second embodiment) is somewhat large If it is large, the process proceeds to step S13, and if it is not large, the process proceeds to step S14.
  • “large to some extent” means that the size of the variation in shape can be regarded as a high possibility of being a pedestrian or a bicycle, and is set in advance based on actual measurements of the pedestrian or bicycle It is determined by comparing with a threshold.
  • step S13 which proceeds when the variation in shape is large to a certain extent, the image portion Q is determined to be a pedestrian including a bicycle, and the process returns to step S1.
  • the image of the area q occupied by the moving object is a pedestrian, although the variation in shape becomes large as described above, the change in the position of the center of gravity of the area q is small. You may add as conditions.
  • Step S14 and the subsequent steps S15 to S18 when the variation in shape is not large are the same as # 8 to # 12 in the first embodiment, and the process proceeds to step S15 when the enlargement ratio K tends to increase. If the image portion Q is determined to be an approaching vehicle, and the enlargement ratio K tends to increase by the same factor, the process proceeds to step S17, and it is determined to be a parallel traveling vehicle. If the enlargement ratio K is not equal to any increase At step S18, it is determined that the vehicle is a distant vehicle (non-approaching object).
  • the vehicle 200 on which the approaching object detection system 100b is mounted is about to enter an intersection.
  • the vehicle 200 on which the approaching object detection system 100b is mounted is about to enter an intersection.
  • the approaching object detection system 100b starts object detection processing, and first acquires vehicle information (step S1). Then, based on the vehicle information, when entering the intersection in this forward traveling, the left and right landscapes Jm and Jn are photographed by the cameras 10 and 20 mounted at the front end of the vehicle (step S2). The images P captured by the cameras 10 and 20 are acquired as images P continuous in time series (step S3).
  • the image processing unit 210 when the vehicle 200 is moving without stopping based on the vehicle information, the stopped object in the current image P (t) based on the shift information and the vehicle speed information A size correction is performed to correct the frame magnification so that the projection size of (background) and the projection size of the stop object (background) in the immediately preceding image P (t-1) are the same. It corrects so that the change of the size of the stop object image in the image P resulting from traveling does not occur.
  • lens distortion correction is also performed to correct the image in the image P (t) to be an image without distortion (step S4). In addition, when the own vehicle 200 has stopped, this size correction is not performed.
  • the first moving area detection means 30b performs optical flow processing based on the images P (t), P (t + 1), P (t + 2),. ), P (t + 1),..., Areas q (t), q (t + 1), q (t + 2) occupied by the moving object moved in the horizontal direction (specific direction) in time series are detected (step S5) . Further, the variation calculation unit 220 of the first moving region detection means 30b obtains the variation of the outer shape in the region q occupied by the moving body, and the variation of the area in the second embodiment (step S6). Further, the first moving area detecting means 30b sets rectangular image portions Q (t), Q (t + 1), Q (t + 2),... Including the area q occupied by the moving body (step S7).
  • the second moving area detection means 40 extracts the luminance in the vertical direction in the image portion Q (step S8), and further associates the horizontal scale and the position in the image portion Q of the image P that precedes and follows the image. Then (step S9), the enlargement factor K of the image portion Q is calculated (step S10).
  • the object indicated by the image portion Q is another one based on the enlargement factor K detected by the second moving area detection means 40 and the variation calculated by the variation calculation unit 220. It is determined whether the vehicle 300 is a pedestrian, a pedestrian, a parallel traveling vehicle, or a distant vehicle.
  • the size of the area q occupied by the moving object determined to have movement by the optical flow is somewhat large and the variation of the shape is large to a certain extent. Since it is not constant, it is determined as a pedestrian including a bicycle (S11 ⁇ S12 ⁇ S13).
  • the enlargement ratio K in the area q determined to have movement by the optical flow tends to increase, it is determined to be another vehicle 300 approaching (S11 ⁇ S12 ⁇ S14 ⁇ S15) and enlarged If the ratio K tends to be equal, it is determined that the vehicle is a parallel traveling vehicle (S11 ⁇ S12 ⁇ S14 ⁇ S16 ⁇ S16 ⁇ S17).
  • those for which the enlargement ratio K does not tend to increase and to be equal have a detection target as a distant vehicle as having a low degree of danger to the host vehicle for the time being Remove from things.
  • a camera monitor system or the like is used for the occupant of the host vehicle such as a driver as in the first embodiment. An action to alert the user is performed. In this case, it is preferable to differentiate the display between the approaching vehicle and the pedestrian.
  • the camera 240 at the rear of the vehicle captures a picture of the rear of the vehicle 200, and the same process as described above is performed. To be executed. For example, if you parked in a parking lot, etc. instead of moving backwards, when you leave the parking lot, you will be moving backwards, and it will be difficult to visually recognize the rear of the vehicle, but as mentioned above In the approaching object detection system in which the camera is fixed, the system reliably detects an approaching object approaching to the moving vehicle even when the moving object is difficult to visually recognize. Therefore, the driving operation of the vehicle can be safely supported.
  • the following effects a) b) c) d) can be obtained as in the first embodiment.
  • a) Optical flow processing in which only the velocity vector in the substantially horizontal direction is detected when the first moving area detection means 30 b extracts an object with motion from two images P and P that are sequentially linked in time series As compared with detection with velocity vectors in all directions, it is possible to accurately detect (extract) an object moving in a substantially horizontal direction in a short time while reducing the calculation processing load.
  • the second moving area detection means 40 obtains the distribution (profile) of signal values along the vertical direction for each image portion Q, Q. Distribution of signal values of the image portion Q (t) in the previous image P (t) in time series and distribution of signal values of the image portion Q (t + 1) in the image P (t + 1) after time series
  • the enlargement ratio K is determined by making the signal values correspond to each other. Thus, based on the enlargement factor K, it can be easily determined whether the moving object is an object that is truly approaching.
  • the special effects of the second embodiment of the following e) to f) can be achieved.
  • e) In order to detect an approaching object in the first movement area detection means 30b, when the host vehicle 200 is in the non-stop state, the enlargement ratio of the image P is set based on the vehicle speed and the traveling direction of the host vehicle 200. , Correction was made in the direction in which the size of the non-moving object (background) does not change. For this reason, it is possible to prevent erroneous detection due to the size change of the non-moving object (background) on the image P due to the forward and backward movement of the vehicle 200.
  • the optical flow is limited to a specific direction, and the size change of the non-moving object image (background) on the image P due to the movement of the own vehicle 200 is a method of lightening the operation processing load using dynamic programming. Can be removed to improve the detection accuracy of the approaching object. Moreover, since the removal of the size change of the non-moving object is performed by changing the enlargement factor of the image P, the load of the correction processing is compared to performing the image processing of the specific image in the image P. Can be small.
  • the detection direction of the optical flow is specified, and the dynamic programming method A pedestrian can be detected with high accuracy even with a means with low computational processing load using.
  • the moving direction is unstable or the moving speed is slow, so that, for example, optical flow is difficult to appear between the images P adjacent to each other in time series, and the enlargement ratio K is also low.
  • determination of an approaching object is performed even when the host vehicle 200 is moving, and when there is a pedestrian approaching the host vehicle 200, the host vehicle 200 is stopped.
  • the probability of contact with a pedestrian is higher than when it is. Therefore, it is more effective to perform pedestrian determination as an approaching object in this manner.
  • the pedestrian is determined by narrowing the size of the area q occupied by the moving object to a certain extent and the pedestrian with a low possibility of contact is excluded from the determination. Can reduce the load on
  • the variation is determined based on the numerical value accumulated over a predetermined time for the difference in the area of the region q in the preceding and succeeding images P.
  • the processing is based on the area, the comparison is simple and the calculation can be simplified.
  • the area difference is accumulated and determined, even if the discrimination is difficult between the images P adjacent to each other like the pedestrian by the optical flow and the dynamic programming, the pedestrian's It becomes possible to judge.
  • the cameras 10, 20, and 240 are fixed to the front and rear of the vehicle 200, but in the present invention, the installation positions of the cameras are limited to these.
  • the camera in the approaching object detection system according to the present invention may be fixed to any part of the vehicle, and may be provided, for example, on the side of the vehicle.
  • the camera is not limited to one fixed to a vehicle, and may be attached to a structure or the like.
  • the size of the image of the non-moving object (background) in the image P is corrected to be the same when the vehicle 200 is moving.
  • the size of the images may not be completely the same as long as correction is performed in the direction in which the change in the size of the image is suppressed. Also in this case, it is possible to suppress false detection, as compared with the case where correction is not performed.
  • the size of the stop object becomes the same on the images that sequentially go back and forth in time series according to the back and forth movement amount of the vehicle.
  • the size correction is performed, but the following correction may be added. That is, in the image processing unit, when the image on the image P moves in the horizontal direction as the vehicle turns, the image of the non-moving object stays at the same place on the image P according to the steering angle and the vehicle speed. A correction may be made to move the center of the image P in the horizontal direction (x-axis direction).

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Image Analysis (AREA)
  • Traffic Control Systems (AREA)
  • Closed-Circuit Television Systems (AREA)
  • Image Processing (AREA)
PCT/JP2009/069649 2008-11-19 2009-11-19 接近物体検知システム Ceased WO2010058821A1 (ja)

Priority Applications (3)

Application Number Priority Date Filing Date Title
US13/129,869 US8712097B2 (en) 2008-11-19 2009-11-19 Approaching object detection system
EP09827602.5A EP2369552B1 (en) 2008-11-19 2009-11-19 Approaching object detection system
CN2009801458033A CN102257533B (zh) 2008-11-19 2009-11-19 接近物体检测系统

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
JP2008296030 2008-11-19
JP2008-296030 2008-11-19
JP2009255666A JP5421072B2 (ja) 2008-11-19 2009-11-09 接近物体検知システム
JP2009-255666 2009-11-09

Publications (1)

Publication Number Publication Date
WO2010058821A1 true WO2010058821A1 (ja) 2010-05-27

Family

ID=42198260

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2009/069649 Ceased WO2010058821A1 (ja) 2008-11-19 2009-11-19 接近物体検知システム

Country Status (5)

Country Link
US (1) US8712097B2 (https=)
EP (1) EP2369552B1 (https=)
JP (1) JP5421072B2 (https=)
CN (1) CN102257533B (https=)
WO (1) WO2010058821A1 (https=)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2541465A3 (en) * 2011-06-27 2017-09-13 Clarion Co., Ltd. Vehicle periphery monitoring system
CN113689714A (zh) * 2020-05-19 2021-11-23 大唐移动通信设备有限公司 通行控制方法、装置及交通控制系统

Families Citing this family (30)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TW201221391A (en) * 2010-11-26 2012-06-01 Hon Hai Prec Ind Co Ltd Vehicle and backing up method thereof
JP5533766B2 (ja) * 2011-04-05 2014-06-25 株式会社デンソー 車両用表示装置
JP5776769B2 (ja) * 2011-05-24 2015-09-09 富士通株式会社 物体接近検知装置、物体接近検知方法、及びプログラム
JP5868426B2 (ja) * 2011-12-13 2016-02-24 株式会社日立製作所 停立人物の向き推定方法
JP5792091B2 (ja) * 2012-02-16 2015-10-07 富士通テン株式会社 物体検出装置及び物体検出方法
JP5615862B2 (ja) * 2012-03-07 2014-10-29 クラリオン株式会社 車両周囲監視装置
JP5192597B1 (ja) * 2012-04-03 2013-05-08 株式会社ナナオ 動き画像領域判定装置またはその方法
JP5867273B2 (ja) * 2012-04-27 2016-02-24 富士通株式会社 接近物体検知装置、接近物体検知方法及び接近物体検知用コンピュータプログラム
WO2014017600A1 (ja) * 2012-07-27 2014-01-30 日産自動車株式会社 立体物検出装置および立体物検出方法
WO2014019132A1 (en) * 2012-07-31 2014-02-06 Harman International Industries, Incorporated System and method for detecting obstacles using a single camera
EP2701093B1 (en) * 2012-08-20 2016-06-22 Honda Research Institute Europe GmbH Sensing system and method for detecting moving objects
JP2014131257A (ja) * 2012-11-27 2014-07-10 Ricoh Co Ltd 画像補正システム、画像補正方法及びプログラム
EP2757527B1 (en) 2013-01-16 2018-12-12 Honda Research Institute Europe GmbH System and method for distorted camera image correction
JP2014044730A (ja) * 2013-09-24 2014-03-13 Clarion Co Ltd 画像処理装置
US9248832B2 (en) * 2014-01-30 2016-02-02 Mobileye Vision Technologies Ltd. Systems and methods for detecting traffic signal details
JP6496982B2 (ja) * 2014-04-11 2019-04-10 株式会社デンソー 認知支援システム
US9390348B2 (en) * 2014-05-19 2016-07-12 Jinling Institute Of Technology Method for categorizing objects in image
JP6323262B2 (ja) * 2014-08-29 2018-05-16 トヨタ自動車株式会社 車両の接近物体検出装置
EP2990290B1 (en) * 2014-09-01 2019-11-06 Honda Research Institute Europe GmbH Method and system for post-collision manoeuvre planning and vehicle equipped with such system
KR20160075135A (ko) * 2014-12-19 2016-06-29 현대모비스 주식회사 자동차 물체인식 시스템 및 그 동작방법
EP3282392B1 (en) * 2015-08-28 2021-09-29 Veoneer Sweden AB Vision system and method for a motor vehicle
JP6620977B2 (ja) * 2015-10-09 2019-12-18 パナソニックＩｐマネジメント株式会社 表示制御装置、投影装置、および表示制御プログラム
EP3376468B1 (en) * 2015-12-02 2020-01-29 Mitsubishi Electric Corporation Object detection device and object detection method
CN108122253A (zh) * 2017-12-04 2018-06-05 广东欧珀移动通信有限公司 接近检测方法、装置及电子设备
DE102018208278A1 (de) * 2018-05-25 2019-11-28 Robert Bosch Gmbh Betriebsassistenzverfahren, Steuereinheit, Betriebsassistenzsystem und Arbeitsvorrichtung
CN113811931B (zh) 2019-06-20 2024-02-02 住友电气工业株式会社 车载通信系统、交换机装置、功能部、通信控制方法以及通信控制程序
US11228737B2 (en) * 2019-07-31 2022-01-18 Ricoh Company, Ltd. Output control apparatus, display terminal, remote control system, control method, and non-transitory computer-readable medium
JP7236556B2 (ja) * 2019-10-14 2023-03-09 株式会社デンソー 物体検知装置および物体検知プログラム
US11783490B2 (en) 2020-03-19 2023-10-10 Objectvideo Labs, Llc Ground plane filtering of video events
JP7635116B2 (ja) * 2021-02-08 2025-02-25 フォルシアクラリオン・エレクトロニクス株式会社 外界認識装置

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000074645A (ja) * 1998-08-27 2000-03-14 Yazaki Corp 周辺監視装置及び方法
JP2004056763A (ja) 2002-05-09 2004-02-19 Matsushita Electric Ind Co Ltd 監視装置、監視方法および監視用プログラム
JP2007233469A (ja) 2006-02-27 2007-09-13 Toshiba Corp 物体検出装置及びその方法
JP2007257025A (ja) 2006-03-20 2007-10-04 Clarion Co Ltd 移動体監視装置

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6081606A (en) * 1996-06-17 2000-06-27 Sarnoff Corporation Apparatus and a method for detecting motion within an image sequence
US6342904B1 (en) * 1998-12-17 2002-01-29 Newstakes, Inc. Creating a slide presentation from full motion video
JP3808242B2 (ja) * 1999-07-26 2006-08-09 パイオニア株式会社 画像処理装置、画像処理方法及びナビゲーション装置
JP2002314989A (ja) 2001-04-10 2002-10-25 Yazaki Corp 車両用周辺監視装置
US7266220B2 (en) 2002-05-09 2007-09-04 Matsushita Electric Industrial Co., Ltd. Monitoring device, monitoring method and program for monitoring
DE102005013920B4 (de) * 2004-03-26 2007-12-13 Mitsubishi Jidosha Kogyo K.K. Frontsicht-Überwachungsvorrichtung
US20060088188A1 (en) * 2004-10-11 2006-04-27 Alexander Ioffe Method for the detection of an obstacle
JP2008219063A (ja) * 2007-02-28 2008-09-18 Sanyo Electric Co Ltd 車両周辺監視装置及び方法
US8300887B2 (en) * 2007-05-10 2012-10-30 Honda Motor Co., Ltd. Object detection apparatus, object detection method and object detection program
CN100568266C (zh) * 2008-02-25 2009-12-09 北京理工大学 一种基于运动场局部统计特征分析的异常行为检测方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000074645A (ja) * 1998-08-27 2000-03-14 Yazaki Corp 周辺監視装置及び方法
JP2004056763A (ja) 2002-05-09 2004-02-19 Matsushita Electric Ind Co Ltd 監視装置、監視方法および監視用プログラム
JP2007233469A (ja) 2006-02-27 2007-09-13 Toshiba Corp 物体検出装置及びその方法
JP2007257025A (ja) 2006-03-20 2007-10-04 Clarion Co Ltd 移動体監視装置

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
MASAMI MIZUTANI ET AL.: "Gazo Ninshiki Gijutsu no Driver Shikaku Shien eno Oyo", FUJITSU, vol. 59, no. 4, 10 July 2008 (2008-07-10), pages 397 - 402, XP008169038 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2541465A3 (en) * 2011-06-27 2017-09-13 Clarion Co., Ltd. Vehicle periphery monitoring system
CN113689714A (zh) * 2020-05-19 2021-11-23 大唐移动通信设备有限公司 通行控制方法、装置及交通控制系统

Also Published As

Publication number Publication date
EP2369552A4 (en) 2016-08-17
US20110228985A1 (en) 2011-09-22
CN102257533A (zh) 2011-11-23
JP5421072B2 (ja) 2014-02-19
JP2010152873A (ja) 2010-07-08
US8712097B2 (en) 2014-04-29
EP2369552B1 (en) 2018-09-19
CN102257533B (zh) 2013-11-13
EP2369552A1 (en) 2011-09-28

Similar Documents

Publication Publication Date Title
WO2010058821A1 (ja) 接近物体検知システム
US8174375B2 (en) Detection system for assisting a driver when driving a vehicle using a plurality of image capturing devices
EP2933790B1 (en) Moving object location/attitude angle estimation device and moving object location/attitude angle estimation method
JP3807331B2 (ja) カメラの汚れ検出装置およびカメラの汚れ検出方法
JP5054612B2 (ja) 接近物検出装置および接近物検出方法
US20150062141A1 (en) Alert display device and alert display method
JP2005309797A (ja) 歩行者警報装置
US20180330619A1 (en) Display device and display method for displaying pictures, and storage medium
JP2010044561A (ja) 乗物搭載用監視装置
US12101580B2 (en) Display control apparatus, display control method, and program
US12165418B2 (en) Object detection apparatus, object detection method, and object detection program
US9827906B2 (en) Image processing apparatus
KR100355993B1 (ko) 차선 이탈 경보장치 및 방법
JP2005309660A (ja) 車両用右左折支援装置
JP4826355B2 (ja) 車両周囲表示装置
US11410288B2 (en) Image processing apparatus
JP2018073049A (ja) 画像認識装置、画像認識システム、及び画像認識方法
US12361721B2 (en) Image processing apparatus for estimating vehicle spatial motion
JP3942289B2 (ja) 車両監視装置
JP2009214790A (ja) 撮像装置および方法、並びに、画像処理装置および方法
JP2000251199A (ja) 車両用後側方監視装置
JP2004334784A (ja) 確認動作検出装置及び警報システム
JP2006107000A (ja) 画像異常判定方法及び画像異常判定装置
JP2017049666A (ja) 飛び出し物体検知装置および飛び出し物体検知方法
JP2021170166A (ja) 画像処理装置、撮像装置、画像処理方法およびプログラム

Legal Events

Date Code Title Description
WWE Wipo information: entry into national phase

Ref document number: 200980145803.3

Country of ref document: CN

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 09827602

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 13129869

Country of ref document: US

NENP Non-entry into the national phase

Ref country code: DE

WWE Wipo information: entry into national phase

Ref document number: 2009827602

Country of ref document: EP