JP2013047878A - Road situation grasping device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a road situation grasping device for accurately determining road situations with respect to a monitoring object lane by preventing the erroneous determination of a vehicle which is traveling in a non-monitoring object lane.SOLUTION: A road situation grasping device picks up an image on a road consisting of a plurality of lanes in which a vehicle is traveling by image pickup means, and determines the situation of the road on the basis of the picked-up image. The road situation grasping device is characterized by creating a moving vector image from the picked-up image, detecting the vehicle which is traveling in a non-monitoring object lane from the directivity of the created moving vector image and the traveling direction information of the vehicle provided in advance, and removing the image of the detected vehicle which is traveling in the non-monitoring object lane from the picked-up image.

Description

  Embodiments described herein relate generally to a road condition grasping apparatus that determines a road condition.

  On a toll road such as an expressway, an image on the road is captured with a camera installed on the roadside, and a vehicle traveling on the road is detected from the captured image by image processing or the like, and the position of the detected vehicle, 2. Description of the Related Art A road condition grasping device that determines current road conditions (road traffic jams, stopped vehicles, low-speed vehicles, etc.) from information such as speed is known.

  In such a road condition grasping device, since it is not assumed that the unmonitored lane is reflected in the image due to the installation position of the camera, a part of the vehicle traveling in the unmonitored lane is projected onto the monitored lane. In such a case, there is a possibility of misjudging road conditions.

  That is, depending on the installation position of the camera, for example, as shown in FIG. 21, the unmonitored lane 2 is reflected in the image of the monitored lane 1 on which the vehicle 3 travels. When a part of the vehicle 4 to be projected is projected onto the monitoring target lane 1, there is a possibility that the road condition is erroneously determined.

JP-A-6-274786

  The problem to be solved by the present invention is to provide a road condition grasping device that can prevent erroneous determination of a vehicle traveling in a non-monitoring target lane and accurately determine the road condition for the monitoring target lane. is there.

  In the road situation grasping device according to the embodiment, an image on a road composed of a plurality of lanes on which a vehicle travels is picked up by an imaging unit, and the road situation is determined based on the picked-up image. A movement vector image is created from the captured image, and a vehicle traveling in the non-monitoring lane is detected from the directionality of the created movement vector image and the vehicle traveling direction information given in advance. An image of a vehicle traveling in a non-monitoring target lane is excluded from the captured image.

The schematic diagram which shows roughly the structure of the road condition grasping | ascertainment apparatus which concerns on embodiment. The flowchart explaining operation | movement of the road condition grasping | ascertainment apparatus which concerns on embodiment. The flowchart explaining operation | movement of the road condition grasping | ascertainment apparatus which concerns on embodiment. The schematic diagram which shows an example of the monitoring object lane image registered beforehand. The schematic diagram which shows an example of the background image registered beforehand. The figure which shows an example of the advancing direction information of the vehicle registered beforehand. The schematic diagram which shows an example of the input captured image. The schematic diagram which shows an example of the input captured image. The schematic diagram which shows an example of the produced movement vector image. The schematic diagram which shows an example of the produced vehicle area image. The schematic diagram which shows an example of the produced vehicle area movement vector image. The schematic diagram which shows an example of the extracted vehicle area | region vector. The schematic diagram which shows the specific example of a content of a vector log | history holding part. The schematic diagram which shows an example of the vehicle area image which excluded the image of the vehicle which drive | works the non-monitoring object lane detected erroneously. The schematic diagram which shows an example of the produced background difference image. The schematic diagram which shows an example of the produced stop vehicle area | region image. The schematic diagram which shows an example of the extracted stop vehicle area | region gravity center. The schematic diagram which shows an example of the vehicle area image which excluded the image of the vehicle which stops in the non-monitoring object lane detected erroneously. The schematic diagram which shows the specific example of a content of a vector log | history holding part. The schematic diagram which shows an example of the advancing direction information registered and the stop vehicle area | region gravity center coordinate. The schematic diagram explaining an example in which the vehicle of a non-monitoring target lane is reflected in the image of a monitoring target lane.

Hereinafter, a road condition grasping device according to an embodiment will be described with reference to the drawings.
FIG. 1 schematically shows a configuration of a road condition grasping apparatus according to the present embodiment. In FIG. 1, a road 11 is a toll road such as a highway, for example, and a monitoring target lane 12 and a non-monitoring target lane 13 are arranged side by side, and a vehicle (for example, a passenger car) 14 is illustrated on the monitoring target lane 12 by an arrow a. It is assumed that the vehicle (for example, a truck) 15 travels in the direction of the arrow b (the direction opposite to the monitoring target lane 12) in the non-monitoring target lane 13. It is assumed that the monitored lane 12 and the non-monitored lane 13 are curved as illustrated.

  A video camera (hereinafter simply abbreviated as a camera) 16 as an image pickup means is installed on the side of the road 11 near the curved portion. The camera 16 always captures an image in a predetermined area including the vehicle 14 passing through the curved portion of the monitoring target lane 12 at a constant interval (for example, every 33 ms) and outputs it as a frame image. The output frame image Are sequentially sent to the processing unit 17.

  The processing device 17 detects a vehicle by performing predetermined image processing on an image picked up by the camera 16, and determines a situation such as a traffic event and an abnormal vehicle in the monitored lane 12 from the vehicle detection result. An image processing unit 18 that executes image processing, a vector history holding unit 19 that holds movement vector information obtained by image processing, a parameter holding unit 20 that holds various parameters used for image processing, and a final road from the image processing result It is comprised by the determination part 21 which performs the determination of a condition.

  Here, the road conditions determined by the determination unit 21 are, for example, four states of “road congestion”, “stop vehicle”, “low speed vehicle”, and “normal state”. “Road congestion” means that the monitored lane 12 is congested in the vehicle 14, “stopped vehicle” means that there is a stopped vehicle 14 on the monitored lane 12, and “low speed vehicle” means monitoring It is assumed that the vehicle 14 running on the target lane 12 at a speed equal to or lower than a predetermined value is present. If the road condition is not the above four states, the “normal state” is set.

  Next, the operation of the road condition grasping apparatus according to the present embodiment will be described with reference to the flowcharts shown in FIGS. 2 and 3 in such a configuration.

  First, for example, a monitoring target lane image as shown in FIG. 4 (an image showing the monitoring target lane 12), a background image as shown in FIG. And the traveling direction information of the vehicle (the traveling direction information of the vehicle 14 traveling in the monitored lane 12) as shown in FIG. 6 is held (registered) (step S1).

  Note that the traveling direction information holds, for example, a traveling direction with 0 ° directly above as shown in FIG. 6 (a), and a specific example thereof is shown in FIG. 6 (b). The example of FIG. 6B is a case where the traveling direction information is “150”.

  The camera 16 always captures images in a predetermined area set on the monitored lane 12 at regular intervals, and sequentially sends the captured frame images to the image processing unit 18 of the processing device 17 (step S2).

  The image processing unit 18 creates a movement vector image by image processing from a plurality of frame images taken in time series by the camera 16 (step S3). That is, for example, a movement vector between frames is obtained from a frame image at time t and a frame image at time t + Δt as shown in FIGS. 7 and 8, and is generated as a movement vector image (see FIG. 9). In FIG. 9, the portion 14 a corresponds to the vehicle 14, and the portion 15 a corresponds to the vehicle 15.

  The range created by the movement vector image is an area defined by the monitoring target lane image registered in step S1. Although there are various methods for creating the movement vector image, here, the light and shade pattern matching method is used.

  Next, the image processing unit 18 performs a binarization process based on the presence or absence of a movement vector on the movement vector image of FIG. 9 created in step S3, thereby extracting pixels having a movement vector, and further performing a labeling process. To create a moving vehicle area image (see FIG. 10) (step S4). In FIG. 10, the portion 14 b corresponds to the vehicle 14, and the portion 15 b corresponds to the vehicle 15.

  Next, the image processing unit 18 projects the movement vector image shown in FIG. 9 created in step S3 on the moving vehicle area image shown in FIG. 10 created in step S4, so that the vehicle area label as shown in FIG. Each movement vector image (vehicle area movement vector image) is created (step S5). In FIG. 11, the part 14 c corresponds to the vehicle 14, and the part 15 c corresponds to the vehicle 15.

  Next, the image processing unit 18 performs an average value calculation and a centroid calculation of each pixel for each vehicle region label on the vehicle region movement vector image of FIG. 11 created in step S5, and finally the vehicle Extracted as a region vector (see FIG. 12) (step S6). In FIG. 12, a portion 14d corresponds to the vehicle 14, a portion 15d corresponds to the vehicle 15, and 14e and 15e indicate respective center of gravity points.

  Next, the image processing unit 18 holds (registers) the vehicle area vector of FIG. 12 and the coordinate value (X, Y) of the barycentric point extracted in step S6 in the vector history holding unit 19 in time series (step S7). . FIG. 13 shows a specific content example of the vector history holding unit 19.

  Next, the image processing unit 18 compares the vehicle area vector registered in step S7 with the traveling direction information registered in step S1 (step S8). If the vehicle area vector is significantly different from the traveling direction information, It is determined that the vehicle 15 in the monitoring target lane 13 has been erroneously detected, and the image 15b corresponding to the vehicle 15 is removed from the moving vehicle region image shown in FIG. 10 (step S9), and the vehicle 14 as shown in FIG. A moving vehicle area image of only the corresponding image 14b is created.

  If it demonstrates based on a specific example, when the vehicle area | region vector registered into the vector log | history holding | maintenance part 19 is the value in the frame A of FIG. 13, these vehicle area | region vectors will be compared with advancing direction information, and a difference will be large. In the case of (for example, 90 or more), it is determined that the vehicle 15 traveling on the non-monitoring target lane 13 is erroneously detected, and the image 15b corresponding to the vehicle 15 is excluded from the moving vehicle region image.

  In step S8, if the difference between the vehicle area vector and the traveling direction information is small, it is determined that there is no false detection, and the process skips step S9 and proceeds to step S10.

  Next, the image processing unit 18 creates a background difference image from the captured image obtained in step S2 and the background image registered in step S1 (step S10). That is, a background difference image as shown in FIG. 15 is created from the image at time t + Δt shown in FIG. 8 and the background image shown in FIG. In FIG. 15, the part 14 f corresponds to the vehicle 14, and the part 15 f corresponds to the vehicle 15.

  Next, the image processing unit 18 obtains a stop vehicle area image as shown in FIG. 16 by obtaining a difference between the background difference image of FIG. 15 created in step S10 and the moving vehicle area image shown in FIG. Create (step S11).

  Next, the image processing unit 18 performs a labeling process on the stopped vehicle region image of FIG. 16 created in step S11, further calculates a centroid for each label, and a centroid point 15g of the stopped vehicle region (see FIG. 17) is extracted (step S12).

  Next, when the center of gravity is extracted in step S12 (that is, in the case where there is a stop vehicle region, step S13), the image processing unit 18 immediately preceding the frame (for example, three frames before the frame). Is referenced from the vector history holding unit 19 (step S14). Here, when the difference between the vehicle area vector nearest to the frame and the traveling direction information registered in step S1 is large (step S15), the center of gravity point of the stopped vehicle area extracted in step S12, and the frame If the difference from the center of gravity of the latest moving vehicle area is small (step S16), it is determined that the vehicle 15 in the non-monitoring target lane 13 is erroneously detected, and the stop shown in FIG. The image 15f corresponding to the vehicle 15 is removed from the vehicle area image, and a stopped vehicle area image as shown in FIG. 18 is created (step S17).

  Explaining based on a specific example, the barycentric coordinates (X, Y) and the vehicle area vector of the moving vehicle area registered in the vector history holding unit 19 are the values in the frame B of FIG. 19 and extracted in step S12. When the barycentric coordinates (X, Y) of the stop vehicle area and the advance direction information registered in advance are the values shown in FIG. 20, the vehicle area vector in FIG. 19 is compared with the advance direction information in FIG. If the difference is small (for example, less than 50, for example), if the difference is small (for example, less than 50). It is determined that the vehicle 15 stopped in the non-monitoring target lane 13 is erroneously detected, and the image 15f corresponding to the vehicle 15 is excluded from the stopped vehicle region image.

  It should be noted that when the center of gravity point is not extracted in step S13 (that is, when there is no stop vehicle region), or when the difference between the vehicle region vector and the traveling direction information is small in step S15, or step S16. If the difference between the center of gravity of the stopped vehicle area and the center of gravity of the moving vehicle area is large, it is determined that there is no false detection, and the process proceeds to step S18.

  Next, the image processing unit 18 sends the moving vehicle region image of FIG. 14 and the stopped vehicle region image of FIG. 18 obtained as described above to the determination unit 21 (step S18). The situation is determined (step S19).

  Hereinafter, an example of the road condition determination process performed by the determination unit 21 will be briefly described. For example, it is possible to recognize from the moving vehicle area image whether the vehicle 14 traveling in the monitoring target lane 12 is detected, and from the stopped vehicle area image, whether the vehicle 14 stopped in the monitoring target lane 12 is detected. By converting the coordinate value of the vehicle area into an absolute distance using a conversion table previously stored in a memory, the position, vehicle area, and traveling speed of the detected vehicle 14 (area) are obtained.

  Next, based on the vehicle information detected in this manner, at least four conditions of road congestion, a stopped vehicle, a low-speed vehicle, and a normal state are determined as the status of the monitored lane 12. Specifically, for example, it is determined whether or not the monitored lane 12 is congested based on the position, vehicle area, and traveling speed of the vehicle 14 acquired as described above, and the acquired traveling speed is determined. Based on this, a stop vehicle and a low-speed vehicle are determined.

  The details of such road condition determination processing are described in Japanese Patent Application No. 2008-207034, for example.

  The operation described above is for one frame image, and thus the above operation is repeated for each frame image.

  According to the road condition grasping device of at least one embodiment described above, even if the vehicle 15 in the non-monitoring target lane 13 is erroneously detected, it can be excluded. Thereby, it is possible to prevent erroneous determination of the vehicle 15 traveling in the non-monitoring target lane 13 and to accurately determine the road condition with respect to the monitoring target lane 12.

  Although several embodiments of the present invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and their modifications are included in the scope and gist of the invention, and are also included in the invention described in the claims and the equivalents thereof.

  DESCRIPTION OF SYMBOLS 11 ... Road, 12 ... Monitoring target lane, 13 ... Non-monitoring target lane, 14, 15 ... Vehicle, 16 ... Video camera (imaging means), 17 ... Processing apparatus, 18 ... Image processing part, 19 ... Vector history holding part, 20... Parameter holding unit, 21.

Claims (4)

  In a road situation grasping device that takes an image on a road composed of a plurality of lanes in which a vehicle travels and picks up an image on the road and determines the road situation based on the taken image, a movement vector image is obtained from the taken image. A vehicle that travels in the non-monitoring target lane is detected from the directionality of the generated moving vector image and the vehicle traveling direction information given in advance, and an image of the vehicle that travels in the detected non-monitoring target lane Is excluded from the picked-up image. On a road in which a monitoring target lane and a non-monitoring target lane are arranged side by side, an image on the monitoring target lane is captured by an imaging unit, and the captured image is subjected to predetermined image processing to thereby perform the monitoring target lane. A road condition grasping device for judging the situation of
Movement vector image creation means for creating a movement vector image from the captured image;
Vehicle detection means for detecting a vehicle traveling in the non-monitoring lane from the directionality of the movement vector image created by the movement vector image creation means and the traveling direction information of the vehicle given in advance;
Image excluding means for excluding the image of the vehicle traveling in the non-monitoring target lane detected by the vehicle detecting means from the captured image;
A road condition grasping device characterized by comprising:   In a road situation grasping device that takes an image on a road composed of a plurality of lanes in which a vehicle travels and picks up an image on the road and determines the road situation based on the taken image, a movement vector image is obtained from the taken image. A stop vehicle region image in which the vehicle is stopped is generated from the captured image, and the center of gravity coordinates of the generated stop vehicle region image and the generated movement vector image closest to the stop vehicle region image are generated. The vehicle that stops in the non-monitoring target lane is detected from the directionality and the center-of-gravity coordinates of the vehicle and the traveling direction information of the vehicle given in advance, and the captured image of the vehicle that stops in the non-monitoring target lane is detected. Road condition grasping device characterized by excluding from On a road in which a monitoring target lane and a non-monitoring target lane are arranged side by side, an image on the monitoring target lane is captured by an imaging unit, and the captured image is subjected to predetermined image processing to thereby perform the monitoring target lane. A road condition grasping device for judging the situation of
Movement vector image creation means for creating a movement vector image from the captured image;
A stopped vehicle area image creating means for creating a stopped vehicle area image in which the vehicle is stopped from the captured image;
From the barycentric coordinates of the stopped vehicle area image extracted by the stopped vehicle area image creating means, the directionality and barycentric coordinates of the created movement vector image nearest to the stopped vehicle area image, and the advance direction information of the vehicle given in advance Vehicle detection means for detecting a vehicle that stops in the unmonitored lane;
Image excluding means for excluding the image of the vehicle that stops in the non-monitoring target lane detected by the vehicle detecting means from the captured image;
A road condition grasping device characterized by comprising:

Images