JP2010185824A - Vehicle detection device, and method and program of the same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a vehicle detection device for installing a plurality of sensors on a road or a side surface of the road and accurately detecting a vehicle. <P>SOLUTION: This vehicle detection device for merging the plurality of sensors and detecting the vehicle uses a first sensor suitable for measuring the shape of an object and a second sensor suitable for measuring the distance to the object, refers to a rejection range table having a rejection range for disabling the detection result of the vehicle before and after the vehicle position detected by the second sensor, determines whether the detection result by the first sensor can be employed, and merges and outputs the detection results by the first sensor and second sensor. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an object detection technique for detecting an object such as a vehicle installed on a road or a side surface of a road.

  Conventionally, a fusion sensor that improves the detection accuracy of an object by utilizing a plurality of sensors has been performed. Patent Document 1 proposes a method of calculating the existence probability density from the vehicle presence prediction area of each sensor and the detection information of the sensor, and determining the position of the vehicle.

  However, in the above method, although a method that takes into account the characteristics of the resolution and measurement range is disclosed regarding the correlation processing of a plurality of detection information, no consideration has been given to erroneous detection patterns of individual sensors. . Undetected and false detections that occur in real sensors are not only simple spatial characteristics such as resolution, but also low undetected but many false detections, many undetected but few false detections, etc. This is also caused by the difference in the detection characteristics of the sensors.

  As described above, when sensors having greatly different characteristics of undetected and erroneous detection are used and the detection results are fused, there is a problem that it is difficult to obtain the fusion effect.

JP 2000-99875 A

  In order to solve the above-described problem, the present invention combines the two detection units of the first detection unit with few undetections but many false detections and the second detection unit with few false detections but many undetections, By setting an area in which the detection result by the first detection unit is negated around the detection position by the second detection unit, highly accurate vehicle detection is realized.

  One aspect of the invention is a vehicle detection device that detects a vehicle using a plurality of sensors, and detects the presence or absence of the vehicle from data acquired by a first sensor suitable for measuring the shape of an object. A first detection unit, a second detection unit that detects the presence or absence of a vehicle from data acquired by a second sensor suitable for measuring the distance from the object, and the second detection unit Rejection determination unit that refers to a rejection range table in which a rejection range that invalidates vehicle detection results before and after the vehicle position is set as a reference and determines whether or not the detection result by the first detection unit can be adopted. And a fusion processing unit that fuses and outputs the detection result of the first detection unit and the detection result of the second detection unit based on the determination result of the determination unit. The present invention relates to a detection device.

  That is, the vehicle detection apparatus according to the present invention detects a vehicle using a sensor such as a visible camera, which is suitable for measuring the shape of an object (for example, there are few undetected but many false detections). A second detection unit suitable for measuring the distance between one detection unit and an object, for example, using a sensor such as a millimeter wave radar to detect a vehicle (few false detections but many undetections) With the vehicle position detected by the second detection unit as a reference, according to the rejection range set before and after the vehicle position, the first detection unit By adopting a configuration in which the detection result and the search result by the second detection unit are fused, vehicle detection with high accuracy is possible.

  As described above, the first detection unit using a sensor such as a visible camera suitable for measuring the shape of an object and the second detection using a sensor such as a millimeter wave radar suitable for measuring a distance from the object. The vehicle detection apparatus according to the present invention for detecting a vehicle by combining parts sets an area in which no vehicle is present in front and rear with reference to the detection position of the vehicle by the second detection unit, and the first existing in the area. By adopting a configuration for filtering the detection result of the detection unit, it is possible to greatly improve the detection accuracy compared to a conventional fusion sensor using a plurality of sensors.

It is a figure which shows the basic composition of the vehicle detection apparatus which consists of a some detection part which becomes embodiment of this invention. It is a figure which shows the structural example of the rejection determination part which determines rejection of the detection result of the vehicle which becomes embodiment of this invention. It is a figure explaining the relationship between the vehicle detection position and front vehicle which become embodiment of this invention. It is a figure explaining the relationship between the vehicle detection position which becomes embodiment of this invention, and the blind spot range of a vehicle. It is a figure explaining the relationship between the vehicle detection position which becomes embodiment of this invention, and the blind spot range of the vehicle from which a magnitude | size differs. It is a figure which shows the example of a data structure of the rejection range table which becomes embodiment of this invention. It is a figure which shows the operation | movement flow of the vehicle detection in the vehicle detection apparatus which becomes embodiment of this invention. It is a figure which shows the example of data of the vehicle detection in the vehicle detection apparatus which becomes embodiment of this invention. It is a figure which shows the example of a detection result position at the time of converting the vehicle detection data of FIG. 8 which becomes embodiment of this invention to the same coordinate system. It is a figure which shows the vehicle detection example obtained by the fusion process from the output result of the several detection part which becomes embodiment of this invention.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  In the following embodiment, for example, a visible camera that is installed at a height of several meters at an intersection and includes a road on which the vehicle travels is included in the imaging range and a millimeter wave radar that includes the same area as the visible camera in the detection range, A vehicle detection apparatus is assumed that obtains a final detection result by fusing a vehicle detection result from an image acquired by a visible camera and a vehicle detection result from a reflected wave from an object irradiated with millimeter waves.

  FIG. 1 shows a basic configuration of a vehicle detection apparatus including a plurality of detection units according to an embodiment of the present invention. This example shows a functional block diagram of the vehicle detection device.

  The vehicle detection device includes a first detection unit 1 including an input unit 11 that receives input of data acquired by a visible camera and a vehicle determination unit 12 that processes an image acquired by the visible camera and detects the presence or absence of a vehicle. A second detection unit 2, a detection unit 1, and an input unit 21 that receives input of data acquired by the millimeter wave radar and a vehicle determination unit 22 that receives a reflection signal from the measurement object and detects the presence or absence of the vehicle; A detection result position comparison unit 3 that compares the detection result 1 of the detection unit 2 and a detection position of the detection result 2; a rejection determination unit 4 that deletes a false detection using the detection result of the detection unit 2 as a detection result of the detection unit 1; In addition, the detection unit 1 includes a fusion processing unit 5 that fuses the detection result of the detection unit 1 from which erroneous detection has been removed and the detection result of the detection unit 2 together.

  Further, the vehicle determination unit 12 receives the image data of the visible camera from the input unit 11, extracts the vehicle image by performing inter-frame difference or background difference processing on the image data, and outputs the detection result. To do. The vehicle determination unit 22 receives the reflection signal of the measurement object from the input unit 21, determines the presence or absence of the vehicle from the intensity of the reflection signal and the number of reflection points, and outputs the detection result.

  The detection result position comparison unit 3 converts the vehicle detection position based on the detection result obtained by the vehicle determination unit 12 and the vehicle detection position based on the detection result obtained by the vehicle determination unit 22 into the same coordinate system. Then, the converted positions are compared, and if the two detected positions are in the vicinity, they are merged into one result. By this processing, the data related to the detection position included in the detection result is classified into data detected only by the detection unit 1, data detected only by the detection unit 2, and data detected by both.

  Rejection determination unit 4 uses the remaining two detection results including the results detected by detection unit 2 for those detected only by detection unit 1, and the rejection range defined in advance by rejection determination unit 4. It is determined whether or not there is a false detection with reference to (described later), and those determined to be erroneous detection are deleted.

  As described above, the first detector using a sensor such as a visible camera suitable for measuring the shape of an object and the second detector using a sensor such as a millimeter wave radar suitable for measuring the distance from the object. In the vehicle detection device of the present invention having the detection unit, the first detection unit is set by setting an area in which no vehicle exists in front and rear of the detection position of the vehicle detected by the second detection unit. It is possible to suppress false detection.

  Although not shown in the figure, the vehicle detection device is a computer having a CPU (Central Processing Unit) and a memory, and a vehicle detection program for operating the vehicle detection device is stored in an auxiliary storage device (not shown). At the time of startup, it is expanded in the memory and executed by the CPU.

  FIG. 2 shows a configuration example of a rejection determination unit that determines rejection of a vehicle detection result according to the embodiment of the present invention. FIG. 2 shows a detailed configuration of the rejection determination unit 4 of FIG.

  Rejection determination unit 4 includes a vehicle type determination unit 41 that determines the vehicle type from the detection result of only detection unit 2 and the detection results detected by both detection units 1 and 2, and the vehicle position and speed obtained from vehicle type determination unit 41. A forward rejection determination unit 42 that determines rejection ahead of the vehicle detection position by the detection unit 2 based on vehicle information such as a vehicle type and a forward rejection range acquired from the rejection range table 100 in which a rejection range is set in advance. The rear rejection determination unit 43 is configured to determine rearward rejection based on the vehicle information of the determination unit 41 and the rear rejection range of the rejection range table 100.

  The forward rejection determination unit 42 refers to a predefined rejection range table 100 (details will be described later) for the detection result 1 (image data determined to be a vehicle) by the detection unit 1, and the vehicle information of the vehicle type determination unit 41. If the result falls within the rejection range, the detection result 1 is deleted, and if it falls outside the rejection range, data is output to the backward rejection determination unit 43. Then, the backward rejection determination unit 43 refers to the rejection range table 100 for the detection result 1 determined to be out of the range by the forward rejection determination unit 42, determines the backward rejection range based on the vehicle information of the vehicle type determination unit 41, If the result falls within the rejection range, the detection result 1 is deleted, and if it is outside the rejection range, output processing is performed as the detection result after rejection.

  That is, in the front rejection determination unit 42 and the rear rejection determination unit 43, the detection result by the detection unit 1 is deleted if it is within the rejection range, and only the detection result after being deleted is output.

  As described above, the present invention uses the detection result information (vehicle position / speed / vehicle type) including the detection result 2 by the detection unit 2 as a rule defined by the rejection determination unit 4 for the detection result 1 in the rejection range table 100. Therefore, the detection result 1 of only the detection unit 1 is rejected.

  FIG. 3 is a diagram illustrating the relationship between the vehicle detection position and the preceding vehicle according to the embodiment of the present invention. The figure shows an example in which two sensors, a visible camera that inputs data to the input unit 11 of the detection unit 1 and a millimeter wave radar that inputs data to the input unit 21 of the detection unit 2 are installed on the road. Yes. When the detection result position by the detection unit 2 that detects the vehicle from the data of the millimeter wave radar is used as a reference, the rejection range of the detection result of the vehicle by the detection unit 1 in front of it is represented.

  As described above, the first detection unit 1 includes an input unit 11 that receives data acquired from a visible camera installed on a road, and detects a vehicle from the data received by the input unit 11. Similarly, the second detection unit 2 includes an input unit 21 that receives data acquired from a millimeter wave radar installed at the same position on the road, and the input unit 21 includes a vehicle determination unit 12 that performs processing. The vehicle determination part 22 which performs the detection process of a vehicle from the received data is included.

  When the vehicle position as a result of detection by the detection unit 2 exists at a certain position on the road and the safe inter-vehicle distance with respect to the vehicle speed at that location is taken into consideration, the distance between the vehicle and the inter-vehicle distance is in front of the vehicle position. There should be no other vehicles. Therefore, with reference to the vehicle position of the detection result by the detection unit 2, “vehicle length + inter-vehicle distance” is set as the rejection range for the front.

  Regarding the front vehicle length + inter-vehicle distance range, the vehicle length is the total length of a general vehicle (vehicle CO), and the inter-vehicle distance is the detection result detected by the detection unit 2 (vehicle C1, its detection position is a circle on the front of C1. ) And a value obtained from the speed range (1), both of which are obtained from a rejection range table (described later) in which pre-defined values are stored.

  FIG. 4 is a diagram illustrating the relationship between the vehicle detection position and the vehicle blind spot range according to the embodiment of the present invention. In the figure, the detection positions (millimeter wave reflection surfaces) of the vehicles C2 and C3 detected by the detection unit 2 correspond to the front end portions (bumpers and the like) of each vehicle, and are represented by ◯ marks.

  A range from the projection position of the visible camera on the road surface to the rear of the vehicle to the vehicle detection position (circle) detected by the millimeter wave radar is represented as a rear dead angle range. Similarly, the rear dead angle range is acquired from a later-described rejection range table.

  FIG. 5 is a diagram for explaining the relationship between the vehicle detection position and the blind spot range of vehicles having different sizes according to the embodiment of the present invention. FIG. 5 shows the vehicle's rear dead angle range for vehicles having different sizes, such as a normal vehicle C4 and a large vehicle C5. This shows that the rear blind spot range by the large vehicle C5 is wider than the blind spot range by the ordinary vehicle C4.

  As described above with reference to FIGS. 4 and 5, the visible camera has a blind spot with respect to the size and height of the vehicle, and within the blind spot range behind the leading position (correct detection position) of the vehicle, If there is a detection result of the detection unit 1 using a visible camera, it indicates that it is shifted from the correct position, and it can be said that it is a false detection. Further, this blind spot range becomes wider as the position of the vehicle becomes farther, and becomes wider as the size and height of the vehicle become larger.

  FIG. 6 shows a data configuration example of the rejection range table according to the embodiment of the present invention. As a rejection range table 100 in which a predefined rejection range is stored, (a) shows the data configuration of the forward rejection range table, and (b) shows the data configuration of the backward rejection range table. Rejection range table 100 is held in the storage area of the vehicle detection device.

  The forward rejection range table (a) includes a plurality of records including items of a forward rejection range (vehicle length + inter-vehicle distance) with respect to the vehicle speed. The vehicle length is the total length of a general vehicle, and the inter-vehicle distance is obtained from the vehicle speed, which is the detection result of the millimeter wave radar. For example, the forward rejection range “vehicle length + inter-vehicle distance” for a vehicle speed of 40 km / h is set to 45 m.

  In addition, the rear rejection range table (b) is configured with items of normal, large, special, etc. rear rejection ranges (car type blind spot ranges) with respect to the distance from the mounting position of the visible camera to the detection position of the vehicle. For example, when the distance from the visible camera is 40 m, the rear rejection range is set to 5.0 m for ordinary vehicles, 10.0 m for large vehicles, and 15.0 m for special vehicles.

  FIG. 7 shows an operation flow of vehicle detection in the vehicle detection device according to the embodiment of the present invention. First, in step S11, vehicle detection data is acquired from the detection unit 1 using a visible camera as a sensor and the detection unit 2 using a millimeter wave radar as a sensor. Next, in step S12, the detection positions of detection results 1 and 2 obtained by the sensors are converted to the same coordinate system.

  In step S13, it is determined whether there is image data different from the millimeter wave data. If there is different image data as a result of the determination, in step S14, image data at a position different from the detection result position by the detection unit 2 is extracted.

  Next, in step S15, the rejection range table 100 is referenced, and whether the image data is true or false is determined from “vehicle length + inter-vehicle distance” in the front and from the blind spot range in the rear.

  First, in step S16, it is determined whether or not the vehicle is in the rejection range according to the definition of the forward rejection table (a). If the result of determination is within the range, the detection result is deleted in step S20. On the other hand, if it is out of range, it is determined in step S17 whether or not the image data is in the rejection range with reference to the rear rejection range table (b). If it is within the range, the detection result is deleted in step S20. If it is out of the range, it is determined in step S18 that the vehicle is present, and the processing results are merged.

  In step S19, information on the merged vehicle position is output.

  FIG. 8 shows an example of vehicle detection data in the vehicle detection apparatus according to the embodiment of the present invention. (A) shows the image data of the vehicle by the detection part 1 using a visible camera, and (b) shows the reflected signal from the vehicle by the detection part 2 using a millimeter wave radar. These data (a) and (b) correspond to the data obtained in step S11 of FIG.

  In (a), four points I1, I2, I3, and I4 are captured as vehicles. In (b), millimeter wave reflected signals from the vehicle are obtained at three points M1, M2 and M3.

  FIG. 9 shows an example of a detection result position when the vehicle detection data of FIG. 8 according to the embodiment of the present invention is converted into the same coordinate system. (A), (b) has shown the example of the detection result position when the vehicle detection data by each sensor of FIG. 8 is converted into the same coordinate system in step S12 of FIG.

  And each vehicle detection data is determined as follows by the vehicle determination part 12 and the vehicle determination part 22 (equivalent to step S15-S17 of FIG. 7).

  Regarding each image data of (a), I1 is determined to be a misperception of the shadow of the vehicle, I2 and 13 are determined to be a vehicle, and I4 is determined to be a ceiling portion of the vehicle.

  Further, in the detection result by the millimeter wave radar reflected signal of (b), M1 and M2 are determined to be vehicles, but M3 is determined to be noise because the signal intensity is weak.

  FIG. 10 shows a vehicle detection example obtained by the fusion process from the output results of the plurality of detection units according to the embodiment of the present invention. FIG. 10 shows an output result obtained by fusing the vehicle detection results by the sensors of FIGS. 9A and 9B (corresponding to step S19 in FIG. 7). By filtering the detection result by the visible camera based on the vehicle position detected by the millimeter wave radar with a preset front rejection range (front vehicle length + inter-vehicle distance) and rear rejection range (rear dead angle range), As shown in FIG. 10, a fusion output from which erroneous detection is removed is obtained.

The embodiments of the present invention described above are as shown in the following supplementary notes.
(Appendix 1) A vehicle detection device that detects a vehicle using a plurality of sensors,
A first detector for detecting the presence or absence of a vehicle from data acquired by a first sensor suitable for measuring the shape of the object;
A second detector for detecting the presence or absence of a vehicle from data acquired by a second sensor suitable for measuring the distance to the object;
Based on the vehicle position detected by the second detection unit, with reference to a rejection range table in which a rejection range that invalidates the detection result of the vehicle before and after the vehicle position is referenced, the detection result by the first detection unit A rejection determination unit that determines whether or not to adopt,
A fusion processing unit that fuses and outputs the detection result of the first detection unit and the detection result of the second detection unit based on the determination result of the determination unit;
A vehicle detection device comprising:
(Additional remark 2) The said detection range ahead of the vehicle detection position by a said 2nd detection part is made into the total length of the length of a vehicle, and an object distance, The vehicle detection apparatus of Additional remark 1 characterized by the above-mentioned.
(Supplementary Note 3) The vehicle detection apparatus according to Supplementary Note 1 or 2, wherein the rejection range behind the vehicle detection position by the second detection unit is a blind spot range of the vehicle in the first detection unit. .
(Additional remark 4) The rejection range behind the vehicle detection position by the said 2nd detection part can change the rejection range of the detection result of the said 1st detection part by the magnitude | size of the vehicle which the said 2nd detection part detected. The vehicle detection device according to any one of appendices 1 to 3, characterized in that:
(Supplementary Note 5) Any one of Supplementary Notes 1 to 4, wherein the rejection range behind the vehicle detection position by the second detection unit widens the rejection range of the detection result of the first detection unit as it is farther away. The vehicle detection device according to claim 1.
(Supplementary Note 6) A vehicle detection method in a vehicle detection device that detects a vehicle using a plurality of sensors,
A first detection step of detecting the presence or absence of a vehicle from data acquired by a first sensor suitable for measuring the shape of an object;
A second detection step of detecting the presence or absence of a vehicle from data acquired by a second sensor suitable for measuring the distance to the object;
Based on the vehicle position detected by the second detection unit, with reference to a rejection range table in which a rejection range that invalidates the detection result of the vehicle before and after the vehicle position is referenced, the detection result by the first detection unit Rejection determination step for determining whether or not to adopt,
A fusion processing step of fusing and outputting the detection result of the first detection unit and the detection result of the second detection unit based on the determination result of the determination unit;
A vehicle detection method comprising:
(Supplementary note 7) A vehicle detection program in a vehicle detection device for detecting a vehicle using a plurality of sensors,
On the computer,
A first detection step of detecting the presence or absence of a vehicle from data acquired by a first sensor suitable for measuring the shape of an object;
A second detection step of detecting the presence or absence of a vehicle from data acquired by a second sensor suitable for measuring the distance to the object;
Based on the vehicle position detected by the second detection unit, with reference to a rejection range table in which a rejection range that invalidates the detection result of the vehicle before and after the vehicle position is referenced, the detection result by the first detection unit Rejection determination step for determining whether or not to adopt,
A fusion processing step of fusing and outputting the detection result of the first detection unit and the detection result of the second detection unit based on the determination result of the determination unit;
Vehicle detection program for executing

  A highly accurate moving object detection device including vehicle detection on roads.

DESCRIPTION OF SYMBOLS 1 1st detection part 2 2nd detection part 3 Detection result position comparison part 4 Rejection determination part 5 Fusion processing part 11 1st input part 12 Vehicle determination part 21 2nd input part 22 Vehicle determination part 41 Vehicle type discrimination | determination part 42 forward rejection determination unit 43 backward rejection determination unit 100 rejection range table

Claims (5)

A vehicle detection device that detects a vehicle using a plurality of sensors,
A first detector for detecting the presence or absence of a vehicle from data acquired by a first sensor suitable for measuring the shape of the object;
A second detector for detecting the presence or absence of a vehicle from data acquired by a second sensor suitable for measuring the distance to the object;
Based on the vehicle position detected by the second detection unit, with reference to a rejection range table in which a rejection range that invalidates the detection result of the vehicle before and after the vehicle position is referenced, the detection result by the first detection unit A rejection determination unit that determines whether or not to adopt,
A fusion processing unit that fuses and outputs the detection result of the first detection unit and the detection result of the second detection unit based on the determination result of the determination unit;
A vehicle detection device comprising:   The vehicle detection apparatus according to claim 1, wherein the rejection range in front of the vehicle detection position by the second detection unit is a total length of a vehicle length and an object distance.   The vehicle detection apparatus according to claim 1, wherein the rejection range behind the vehicle detection position by the second detection unit is a blind spot range of the vehicle in the first detection unit. A vehicle detection method in a vehicle detection device that detects a vehicle using a plurality of sensors,
A first detection step of detecting the presence or absence of a vehicle from data acquired by a first sensor suitable for measuring the shape of an object;
A second detection step of detecting the presence or absence of a vehicle from data acquired by a second sensor suitable for measuring the distance to the object;
Based on the vehicle position detected by the second detection unit, with reference to a rejection range table in which a rejection range that invalidates the detection result of the vehicle before and after the vehicle position is referenced, the detection result by the first detection unit Rejection determination step for determining whether or not to adopt,
A fusion processing step of fusing and outputting the detection result of the first detection unit and the detection result of the second detection unit based on the determination result of the determination unit;
A vehicle detection method comprising: A vehicle detection program in a vehicle detection device that detects a vehicle using a plurality of sensors,
On the computer,
A first detection step of detecting the presence or absence of a vehicle from data acquired by a first sensor suitable for measuring the shape of an object;
A second detection step of detecting the presence or absence of a vehicle from data acquired by a second sensor suitable for measuring the distance to the object;
Based on the vehicle position detected by the second detection unit, with reference to a rejection range table in which a rejection range that invalidates the detection result of the vehicle before and after the vehicle position is referenced, the detection result by the first detection unit Rejection determination step for determining whether or not to adopt,
A fusion processing step of fusing and outputting the detection result of the first detection unit and the detection result of the second detection unit based on the determination result of the determination unit;
Vehicle detection program for executing

Images