JP2007164566A - System and device of vehicle sensing for traffic-actuated control - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a system and a device of vehicle sensing for traffic-actuated control which highly precisely detect the presence of a vehicle necessary for accurate traffic-actuated control, and generate a high precision traffic-actuated control signal enabling a signal controller to perform traffic-actuated control at appropriate timing. <P>SOLUTION: The system includes an imaging means for periodically acquiring image data including an area near by a stop line; a data processing means for detecting the vehicle on the basis of the image data; and an output means for outputting a traffic-actuated signal in response to the detection of the vehicle. The processing means infers a vehicle type of the vehicle by measuring a vehicle width of the vehicle in the night, estimates a presence area of the vehicle corresponding to the estimated type of the vehicle, and detects the vehicle. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a vehicle control system and apparatus for sensitive control used for sensitive signal control at signalized intersections.

Conventionally, a sensitive signal control method (hereinafter referred to as “sensitive control”) detects the presence of a vehicle or a two-wheeled vehicle (hereinafter referred to as “vehicle or the like”) traveling in a specified inflow lane. And a function for extending the right turn arrow display of the signal light device by detecting a vehicle or the like existing in the right turn lane or a control method for switching the signal light device from red to blue (hereinafter referred to as “recall sensitive control”), Alternatively, various methods such as a control method having a function of shortening the right turn arrow display when a vehicle or the like does not exist (hereinafter referred to as “right turn sensitive control”) are known. As a sensor used in these sensitive signal control methods, an ultrasonic vehicle sensor has been used.
Further, in this specification, a signal indicating that a vehicle or the like existing in a region subject to recall sensitive control or right turn sensitive control is referred to as a “sensitive signal”, and a region for generating a sensitive signal is referred to as a “sensitive signal”. It will be referred to as a “sensitive area”.

  The ultrasonic vehicle detector has a vehicle detection unit directly above the lane near the stop line. An ultrasonic wave is transmitted from the vehicle detection unit toward the road surface, and an ultrasonic wave reflected upon a detection target (hereinafter simply referred to as “detection target”) such as a vehicle passing the road surface or the corresponding lane is received. The presence of a detection target is detected by measuring the time from transmission to reception of ultrasonic waves.

  On the other hand, an image type vehicle sensor is also used for detecting a vehicle or the like. The image-type vehicle detector takes a picture of the road by installing a camera on the roadside. In the photographed image, a rectangular area (hereinafter referred to as “measurement area”) determined by an arbitrary distance in the lane width of the road surface and the lane direction is set as a measurement range of the detection target. An image including the measurement area is periodically acquired from the camera at regular time intervals. A plurality of pixels (sample points) in a measurement region set in advance with respect to the acquired image at the current time are extracted, and the luminance of the sample points is determined.

  The image data at the current time consisting of the brightness of the sample points is compared with the image data of the measurement area for comparison (hereinafter referred to as “background image data”) stored in advance in the memory, and the detection target is detected from the image data. Extract. Further, a binarized edge portion is extracted as the extracted feature quantity of the detection target. Based on the binarized edge portion, the position and width of the vehicle head (hereinafter referred to as “vehicle head”) to be detected are estimated. Further, a region where a vehicle or the like exists (hereinafter simply referred to as “existing region”) is estimated from vehicle length information associated with the vehicle width. The vehicle head position and the existence area determined based on the image data acquired at the time one cycle before the current time with respect to the existence area, and the vehicle head position determined based on the image data acquired at the current time In addition, the vehicle is detected and the traffic flow is measured by tracking the vehicle or the like by time-sequentially associating with the existence area. The association is generally performed by pattern matching of feature amounts (for example, edges) extracted by image processing.

  In addition, the image type vehicle sensor is also used for sensitive control. Since the presence of the detection target can be detected when the detection target exists in the measurement region of the image acquired from the camera of the image type vehicle sensor, a signal (hereinafter, “ The presence detection signal ") is used as it is as a sensitive signal.

JP-A-5-307695

  However, since the detection area of the vehicle detection unit is as narrow as about 1.2 m in the ultrasonic vehicle detector, the traveling position of the detection target deviates from the detection region, or a transmission wave from the vehicle detection unit is generated when waiting for a signal. There is a problem that a detection omission occurs when it hits the road surface between the vehicle and the like. Furthermore, since it is necessary to install a vehicle detection unit directly above the lane in which the detection target travels, it is necessary to install a long arm from the roadside pillar and install the vehicle detection unit here. Depending on the installation location, construction is difficult, and there is also a problem that the cityscape after installation is damaged.

  Further, when a conventional image type vehicle detector is used for sensitive control, the detected presence detection signal is used as it is as a sensitive signal. For this reason, for example, in the case where only the headlights are captured by the camera at night, the existence area of the vehicle cannot be calculated, and when the headlight deviates from the sensitive area, there is a vehicle in the sensitive area. However, there was a problem that it could not be detected.

An object of the present invention is to provide a vehicle sensing system and device for sensitive control that can accurately detect the presence of a vehicle or the like, which is necessary for accurate sensitive control, in view of the problems of the conventional system and device. There is.
Another object of the present invention is to provide a vehicle sensing system and apparatus for sensitive control that generates a highly accurate sensitive signal so that the signal controller can perform sensitive control at an appropriate timing.

  In order to solve the above-mentioned problem, in response to detection of a vehicle or the like, an imaging means for periodically acquiring image data including an area near the stop line, a data processing means for detecting a vehicle or the like based on the image data, and the like A vehicle sensing system for sensitivity control including an output means for outputting a sensitivity signal, wherein the data processing means is a vehicle width of a vehicle or the like at night (if the vehicle width is not visible, the width from the outside of the headlight to the outside) The vehicle type of the vehicle or the like is estimated by measuring the vehicle, the existence region of the vehicle or the like corresponding to the estimated vehicle type is estimated, and the vehicle or the like is detected.

  As a result, the vehicle's presence area is estimated even at night when the headlight deviates from the sensitive area when the image is dark and the vehicle's headlight is only visible. If so, a sensitive signal can be output.

  Further, the data processing means calculates the existence area of the vehicle or the like by measuring the vehicle width and the vehicle length of the vehicle or the like in the daytime, and detects the vehicle or the like (claim 2). As a result, the existence area of the vehicle or the like can be calculated more accurately.

  Furthermore, there is provided a vehicle sensing device for sensitivity control in which the vehicle sensing system for sensitivity control according to claim 1 or 2 is housed in one casing (claim 3).

  Thus, by accommodating the sensitive control vehicle sensing system according to claim 1 or 2 in a single housing, a sensitive control vehicle sensing device that is easy to install can be realized.

  According to the present invention, it is possible to accurately detect the presence of a vehicle or the like necessary for accurately operating the sensitive control. As a result, an accurate sensitive signal can be generated and transmitted to the signal controller, so that optimum sensitive control can be performed.

  FIG. 1 shows an installation conceptual diagram of a zoom camera-integrated image type vehicle sensor 1 installed for right turn sensitive control. An image type vehicle detector incorporating a zoom camera 2 is provided near the upper part of the signal lamp at the intersection outflow part. The zoom camera 2 is installed such that its field of view overlooks the inflow lane including the stop line 15 at the intersection inflow section. A sensitive area 13 for generating a sensitive signal with respect to an image input from the zoom camera 2 is set as a rectangular area 30 m from the stop line 15.

  FIG. 2 shows an installation conceptual diagram of the zoom type camera-integrated image type vehicle sensor 1 installed for recall sensitive control. As in FIG. 1, an integrated image type vehicle sensor 1 is installed near the upper part of the signal lamp at the intersection outflow part. The field of view of the zoom camera 2 is set so as to overlook the inflow lane including the stop line 15 at the intersection inflow portion. In the video input from the zoom camera 2, the sensitive area 13 is set as a rectangular area of 10 m or more from the stop line 15.

  The sensitive area 13 is set by changing the setting value of the control parameter of the built-in zoom lens 3 after connecting the maintenance console 14 to the integrated image type vehicle sensor 1 and adjusting the depression angle of the camera. A rectangular area within the necessary and sufficient range for sensitive control is set from the stop line 15 at the intersection inflow portion while ensuring the spatial resolution necessary for image processing. The sensitive area 13 is set as a rectangular area using a pointing device such as a mouse connected to the maintenance console 14. When the sensitive area 13 is set, the boundary of the lane and the traveling direction of the vehicle for each lane are set in advance.

  If the maintenance console 14 is used, even if conditions such as the size of the intersection and the installation height and position of the image-type vehicle sensor integrated with the zoom camera are different, the setting value of the zoom lens 3 can be adjusted to easily respond. A visual field capable of setting the sensitive region 13 necessary and sufficient for control can be secured.

  Further, the position and size of the sensitive area 13 necessary and sufficient for the sensitive control can be set using the maintenance console 14. For this reason, conventionally, the timing for performing the sensitivity control was corrected by providing a correction time called the minimum compensation blue time on the signal controller side so as to be able to cope with a case where a delay or deviation occurs in the timing for performing the sensitivity control. That need disappears.

  Further, the non-image processing area can be set using the maintenance console 14. Since the area set as the non-image processing area is determined not to be subject to image processing, the image processing described below is not performed in the non-image processing area.

  By setting the non-image processing area, for example, even when an object that causes brightness fluctuation such as a night warning lamp exists in an area including the sensitive area, the object can be removed as an obstacle. Further, for example, even when a road sign or the like is newly installed and exists in the area including the sensitive area after the camera is installed, it can be set to avoid this as an obstacle in image processing.

  FIG. 3 is a block diagram in the zoom type camera-integrated image type vehicle sensor 1. The main body of the control unit 4 includes a maintenance console I / F 3 for connecting a maintenance console, an image input unit 5 for inputting an image signal acquired from the zoom camera 2, a vehicle detection unit 6, a vehicle tracking unit 7, and presence detection. 8 comprises a response signal output instruction unit 9, a background image update unit 10 and a storage unit 16. The storage unit 16 includes image data processed by the image input unit 5, setting values input from the maintenance console 15, vehicle detection unit 6, vehicle tracking unit 7, presence detection unit 8, sensitive signal output instruction unit 9, background Data processed by the image update unit 10 is stored.

  The zoom type camera-integrated image type vehicle detector 1 further includes an output unit 11 for outputting a sensitivity signal to the signal controller based on a response signal output instruction generated by the response signal output instruction unit 9. . In addition, the power supply unit 12 is also provided.

  An outline of processing performed by the control unit 4 will be described. FIG. 4 shows a flowchart of processing performed by the control unit 4. The processing from S1 to S9 in FIG. 4 is performed periodically. It is assumed that background image data in which no vehicle or the like exists is stored in the storage unit 16 in advance based on a plurality of image data at different times during initialization.

  In the image input unit 5, the sensitive region 13 of the image acquired from the zoom camera 2 is determined from M sample points in the crossing direction of the road and N sample points along the traveling direction of the vehicle or the like. It converts into N coordinate data, and memorize | stores the luminance value corresponding to each coordinate data in the memory | storage part 16 as a sample point image (it shall hereafter be called an input image) (step S1). The interval between the sample points is determined so that the processing can be performed at high speed while maintaining the spatial resolution necessary for image processing. The sample points are arranged at equal intervals on the road surface.

  In the vehicle detection unit 6, for example, binarization processing disclosed in Japanese Patent Laid-Open No. 5-307695 is performed on the image obtained in step S <b> 1. In other words, the road is photographed with a camera, the luminance of a plurality of sample points is determined based on the video information, the spatial differentiation process based on the luminance information of each sample point is performed, and binarized by a predetermined threshold value of the differential signal, Edge portions are extracted (step S2). Next, a difference (hereinafter referred to as background difference) with a background image (an image with only a background without a vehicle) stored in the storage unit 16 in advance is obtained (step S3). For the binarized image obtained in step S2, the head position of the vehicle or the like is determined by a method to be described later, and processing for measuring or estimating the existence region of the vehicle or the like is performed (step S4).

  FIG. 5 is a flowchart showing details of the process of estimating the existence area in step S4. First, using the average luminance value of the background image in the sensitive area 13 as an index, it is determined whether the processing time is daytime or nighttime (step S41). The vehicle width to be detected is measured from the image data obtained in step S2. Specifically, when it is determined that it is daytime in step S41, the vehicle width is measured using, for example, the method disclosed in Japanese Patent Laid-Open No. 5-307695. FIG. 7 shows the concept of the vehicle width measurement method. “Horizontal direction” and “vertical direction” used in the following description refer to directions in the image plane as defined in FIG. A line segment including an edge portion horizontal to the image is extracted by applying a mask corresponding to the width of the vehicle or the like to the binarized image obtained in step S2. Further, the horizontal width of the line segment including the extracted horizontal edge portion is measured. Among the measured horizontal widths including the horizontal edge portion, the maximum width (length A in FIG. 7) is set as the vehicle width (step S42). The measured horizontal edge portion only needs to have a certain length or more with respect to the mask, and there may be a portion where the horizontal edge portion is interrupted.

  Next, the position of the vehicle head is determined. As for the determination method, for example, the position of the vehicle head is determined by the method disclosed in JP-A-5-307695. That is, as a result of the binarization process in step S2, a signal (referred to as “reference 1”) different from the background (referred to as “reference 0”) is obtained only in the edge portion and noise portion of the vehicle or the like. Therefore, a mask corresponding to the width of the vehicle or the like is put on. When the number of “symbol 1” in the mask exceeds a certain threshold, the center of gravity of the distribution of “symbol 1” in the mask is set as the position of the vehicle head candidate point. As a result, there may be a plurality of vehicle head candidate points.

  In order to cope with the case where there are a plurality of vehicle head candidate points, the vehicle detection unit 6 sequentially checks the vehicle head candidate points, and it is assumed that there are n vehicle head candidate points in a region where there is almost one vehicle. First, the first (n = 1) vehicle head candidate point is registered as a vehicle head effective point. Next, a point near the vehicle head candidate point after n = 2 is newly set as a vehicle head effective point. And the vehicle head candidate point which did not become the vehicle head effective point is deleted (step S43).

  Since the vehicle body can be seen in the daytime, the area where the pixels are connected in the binarized data binarized in step S2 is labeled as a lump (referred to as labeling process). For the labeled region, the apparent vehicle length is calculated by measuring the length of the vehicle in the traveling direction from the detected vehicle head position (step S44).

  The existence area is estimated from the apparent vehicle length obtained in step S44 and the vehicle width obtained in step S42 (step S47). An ID number is assigned to the set of the vehicle head position obtained in step S43 and the existence area obtained in step S47. The existence area to which this ID number is assigned is stored in the storage unit 16 as a candidate for a vehicle or the like and registered (step S48).

  Even when it is determined at night in step S41, the process of step S42 described above is performed, and a line segment including an edge portion horizontal to the image is extracted by applying a mask corresponding to the vehicle width of the vehicle or the like. Further, the horizontal width of the line segment including the extracted horizontal edge portion is measured. Of the horizontal widths of the line segment including the measured horizontal edge portion, the largest one (the length of B in FIG. 7) is regarded as the vehicle width. It is determined whether the vehicle is a vehicle or the like based on the measured vehicle width. The aforementioned mask is provided for ordinary vehicles, large vehicles, and two-wheeled vehicles, and the vehicle width can be measured without error even when there is one headlight as in a two-wheeled vehicle.

The vehicle head position is obtained by the same process as in step S43. Thereafter, the vehicle width obtained in step S42 is discriminated as a threshold, and the vehicle type is discriminated (step S45). The correspondence between the vehicle width and the vehicle type at night is, for example, a vehicle width of less than 2 m for a small vehicle and 2 m or more for a large vehicle (a vehicle width of 0.6 m or less for a motorcycle).
Since the vehicle body cannot be seen at night, the vehicle type is determined based on the vehicle width obtained in step S42, and then the length of the vehicle or the like stored in advance in the storage unit 16 is determined according to the determination result in step S45. Selection is made (step S46). The existence area of the vehicle or the like is determined from the vehicle width obtained in step S42 and the length of the vehicle or the like obtained in step S46 (step S47). When the process of step S48 is completed, the process proceeds to step S5 of FIG.

Here, referring back to FIG.
Based on the data obtained in step S4 in the vehicle tracking unit 7, the vehicle head position and the existing area determined at the time one cycle before the current time stored in the storage unit 16 and the vehicle head position at the current time And the existence area are matched using a matching method such as template matching. In template matching, an image region within a certain range in the vicinity of the vehicle head position determined at the time one cycle before is set, and the degree of coincidence with the image at the current time is calculated. Next, an image region in a certain range near the vehicle head position is scanned in the traveling direction of the vehicle, and the degree of coincidence with the image at the current time is sequentially calculated. As a result, the correspondence is performed at the position with the highest degree of coincidence. As a correspondence method, there are methods such as a sum of absolute values of image differences or a normalized correlation. The search range for associating images is determined based on the result of the tracking processing of the vehicle or the like shown in step S5 described later up to one cycle before.

  Next, the vehicle tracking unit 7 obtains the total movement distance by accumulating the movement distances for each cycle tracked from the time when the vehicle or the like enters the sensitive area until the current time. The speed is calculated by dividing the total travel distance of the vehicle head position by the total cycle value calculated by accumulating the cycles for each cycle tracked up to the current time after the vehicle or the like enters the sensitive area. By using this speed, the search range of each vehicle or the like is determined and tracked (step S6).

  Further, in the vehicle tracking unit 7, the direction of the movement vector (that is, the traveling direction of the vehicle or the like that has entered the sensitive area 13) obtained in the process of step S 5 is input from the maintenance console I / F 3 and stored in the storage unit 16. It is determined whether the direction of the trajectory of the movement vector obtained from the recorded tracking result of the vehicle or the like matches the traveling direction (step S6). The determination in step S6 is the degree of coincidence of the traveling direction, that is, the condition that the direction of the trajectory of the movement vector is the traveling direction or the opposite direction, and the length of the trajectory of the movement vector is a certain length (for example, 2 m) or more in the vehicle length direction. It is determined on the condition that it is. When the determination condition of step S6 is satisfied, this target is determined as a detection target because it can be normally tracked (step S7).

  By the processing shown from step S6 to step S7, it is possible to prevent erroneous detection other than the sensitive object. That is, even if a pedestrian or a two-wheeled vehicle crossing a pedestrian crossing is included in the sensitive area 13, the direction of these movement vectors does not coincide with the traveling direction of the vehicle and the like, so that these are prevented from being detected erroneously. be able to.

  The presence detection unit 8 determines the presence / absence of a vehicle or the like in the sensitive area 13 based on the positions of all detection targets correctly tracked in the sensitive area 13 and the existing areas obtained in step S4. To do. Since the vehicle head positions and the existence areas of all the vehicles etc. in the sensitive area 13 can be specified in the process of certification of the vehicles etc. shown in step S7, the presence detection unit 8 is sensitive by the overlap of the sensitive area 13 and the existence areas such as the vehicles. Whether the signal is on or off is determined (step S8). Next, the sensitive signal output instructing unit 9 instructs to output a sensitive signal according to the determination result of step S8, and the output unit 11 outputs a sensitive signal according to the output instruction of the sensitive signal output instructing unit 9 (step S9). .

  Further, in step S9, the vehicle tracking unit 7 determines that the detection target is stopped when the speed of the vehicle or the like obtained in the process of step S5 is zero, and measures the stop duration at this time. Hold.

  FIG. 6 is a flowchart showing details of the processing from step S8 to step S9. In step S8, a running state (including stop and parking) of a vehicle or the like existing in the sensing area 13 is determined, and a response signal generation and a background image update process corresponding to the state are sequentially performed.

  In FIG. 6, the presence detection unit 8 determines the overlapping state of the existence area of each vehicle or the like and the sensitive area 13 (step S <b> 801), and obtains the total number N of the vehicles or the like existing in the sensitive area 13 (step S <b> 802). ). When there is no vehicle or the like in the sensitive area 13 (NO in step S803), the sensitive signal output instructing unit 9 does not instruct the output of the sensitive signal (for example, sets the output instruction flag to OFF) (step In step S823, the background image is updated in step S824. Details of the background image update method will be described later.

  In step S803, the presence detection unit 8 determines whether each vehicle or the like has stopped for N vehicles or the like existing in the sensitive area 13 (step S805). If each vehicle or the like has stopped in the process of step S805, the stop duration is measured (step S806). Next, the stop duration time is compared with a preset threshold value (step S807), and if it exceeds this, it is determined that the vehicle is parked and a flag indicating that a parked vehicle or the like exists (referred to as “parking flag”). Is set to ON (step S808). If the stop continuation time measured in step S806 is less than the threshold value, it is determined that the vehicle is stopped, and a flag indicating that a stopped vehicle or the like exists (hereinafter referred to as a “stop flag”) is turned ON (step S809). . If the vehicle or the like is moving, a flag indicating that the vehicle or the like is moving (hereinafter referred to as a “moving vehicle flag”) is turned ON (step S810).

  A series of processing from step S805 to step S812 is executed for N vehicles or the like existing in the sensitive area 13. Note that the moving vehicle flag, the stop flag, and the parking flag are initialized to OFF in advance before the comparison process in step S805. In the processes of steps S808, S809, and S810, a moving vehicle counter that counts the number of moving vehicles, etc. instead of the moving vehicle flag, the stop flag, and the parking flag, and the number of stopped vehicles are counted. There may be provided a parking counter that counts the number of parked vehicles and parked vehicles.

  Regarding the N vehicles and the like existing in the sensitive area 13 in the process of step S811, the presence of each of the moving vehicle, the stopped car, and the parking in the sensitive area 13 is known. Based on this, two types of combinations of presence / absence of three types of vehicles and the like are respectively obtained as shown in Table 1. The “corresponding step in FIG. 6” in Table 1 represents the symbol and number of the step shown in FIG.

  No. in Table 1 8 is a state in which there is no vehicle. Therefore, in the case of NO in step S803, the processing of step S823 and step S824, that is, the sensitive signal output instruction unit 9 does not instruct the output of the sensitive signal (for example, the output instruction flag is turned OFF). The output unit 11 does not output a sensitive signal (for example, the sensitive signal = OFF). Further, the background image update unit 10 updates the area corresponding to the existence area of the vehicle or the like in the background image. No. in Table 1 7 is a case where there is no parking or the like, and there is no parked vehicle or the like and the vehicle or the like is moving, the processing of step S821 and step S822, that is, the sensitive signal output instructing unit 9 instructs to output the sensitive signal (for example, the output instruction flag is set). The output unit 11 outputs a sensitive signal (for example, the sensitive signal is set to ON). Further, the background image update unit 10 updates the background image including the presence area of the vehicle or the like in the background image. No. in Table 1 4 is when there is parking or the like, but there is no other vehicle or the like, the process shown in steps S817 and S818, that is, the response signal output instruction unit 9 stops the output instruction of the response signal (for example, the output instruction flag is changed to OFF). To do). In this case, the output unit 11 stops the output of the sensitive signal (for example, the sensitive signal = OFF).

  Further, the background image update unit 10 does not update the existence area of the vehicle or the like in the background image. The reason is that if the sensitive signal is turned on for a certain time or more, the signal controller determines that the sensor is abnormal and does not execute the subsequent sensitive control. In the other cases of Table 1, since there are other vehicles, whether or not there is parking or the like, the processing shown in steps S819 and S820, that is, the output of the sensitive signal is instructed by the sensitive signal output instructing unit 9 (for example, When the output instruction flag is set to ON), the output unit 11 outputs a sensitive signal (for example, the sensitive signal = ON). Further, the background image update unit 10 does not update the existence area of the vehicle or the like in the background image.

  If a vehicle or the like newly enters the sensitive area 13 with the vehicle or the like parked, 3, the processing shown in steps S819 and S820, that is, the sensitive signal output instruction unit 9 issues an output instruction of the sensitive signal (for example, the output instruction flag is changed to ON). The output unit 11 outputs the sensitivity signal again (for example, the sensitivity signal = ON). Further, the background image update unit 10 does not update the existence area of the vehicle or the like in the background image.

  Further, even when the vehicle or the like entering the sensitive area 13 travels across the opposite lane because the vehicle or the like has stopped on the road side, the movement obtained by the vehicle tracking unit 7 in the processing from step S6 to step S9 in FIG. If it is determined that the direction of the vector coincides with the traveling direction set via the maintenance console I / F 3 by a threshold or more, No. in Table 1 is obtained. 5, the sensitive signal output instructing unit 9 instructs to output the sensitive signal again, and the output unit 11 outputs the sensitive signal. At this time, since there is a stopped vehicle or the like, the background image update unit 10 does not update the existence area of the vehicle or the like (S820).

  The background image update unit 10 updates the background image. Generally, in the background difference processing for extracting the vehicle body from the luminance difference from the road surface, the background image is updated with the passage of time in order to adapt to the temporal change of sunlight. There is a need. The background image update unit 10 updates the background image using the following exponential smoothing method when there is no vehicle or the like (S824) or when the vehicle or the like is traveling (S822).

The exponential smoothing method is a method of performing an operation represented by the following expression on an image obtained by periodically obtaining an image including the sensitive region 13 from the camera at a certain time.
(Background image of the current cycle) = (1−α) × (Background image at a time before the current cycle) + α × (Input image of the current time) (0 <α <1)

  When a vehicle or the like is parked, an area that is hidden by the vehicle and cannot be seen is generated in the background image. It is not preferable to update the background image in this state. Therefore, when the background image update unit 10 detects parking or the like, that is, when the speed of the vehicle or the like obtained by the vehicle tracking unit 7 in step S5 is zero and the time during which the stop continues is equal to or greater than the threshold value. Applicable when the background image update is interrupted in the presence area of the detected vehicle and the vehicle is started again, that is, when the vehicle speed determined by the vehicle tracking unit 7 in step S5> 0 The update of the background image of the existing area is resumed. In this case, the background image in the region where the vehicle is present is updated in consideration of the vehicle width measured in steps S42 and S45 and the measurement error of the vehicle length measured in step S44. It may be for a predetermined area including the existing area.

  The zoom type camera-integrated image type vehicle sensor 1 according to the present invention is an embodiment of the present invention. In the embodiment of the present invention, a device in which the camera and the control unit are separately separated may be used.

  It should be considered that the embodiments and examples disclosed above are illustrative and non-restrictive in every respect. The scope of the present invention is shown not by the above embodiments and examples but by the scope of claims, and is intended to include all modifications and variations within the meaning and scope equivalent to the scope of claims. .

Installation concept diagram of zoom type camera-integrated image type vehicle detector 1 installed for right turn sensitive control Installation conceptual diagram of a zoom camera-integrated image type vehicle sensor 1 installed for recall sensitive control Block diagram of control unit in zoom type camera-integrated image type vehicle sensor 1 Flowchart of processing performed by the control unit 4 Flowchart of processing for estimating the existence area of a vehicle or the like Flowchart of vehicle parking determination, sensitive signal determination and background image update processing Diagram showing the concept of vehicle width measurement method

Explanation of symbols

1 Zoom-type camera-integrated image-type vehicle detector 2 Zoom-type camera 3 Maintenance console I / F
DESCRIPTION OF SYMBOLS 4 Control part 5 Image input part 6 Vehicle detection part 7 Vehicle tracking part 8 Presence detection part 9 Sensitive signal output instruction | indication part 10 Background image generation / update part 11 Output part 12 Power supply part 13 Sensitive area 14 Maintenance console 15 Stop line

Claims (3)

Imaging means for periodically acquiring image data including a region in the vicinity of the stop line;
Data processing means for detecting a vehicle or the like based on the image data;
Output sensing means for outputting a sensing signal in response to detection of the vehicle or the like, and a vehicle sensing system for sensing control,
The data processing means estimates a vehicle type such as the vehicle by measuring a vehicle width of the vehicle or the like at night,
Estimate the existence area of the vehicle etc. corresponding to the estimated vehicle type,
A vehicle sensing system for sensitive control, wherein the vehicle or the like is detected. The data processing means calculates an existence area of the vehicle or the like by measuring a vehicle width and a vehicle length of the vehicle or the like during the daytime,
The vehicle sensing system for sensitive control according to claim 1, wherein the vehicle or the like is detected. A vehicle sensing device for sensitivity control, wherein the vehicle sensing system for sensitivity control according to claim 1 or 2 is housed in a single casing.

Images