JP5993597B2 - Target moving object detection method - Google Patents

Description

  The present invention relates to a target mobile body detection method for detecting a target mobile body that is a detection target among mobile bodies existing in a detection determination area.

  Conventionally, an optical distance meter such as a laser distance meter is used to irradiate a laser beam into a detection determination area at every scanning angle to detect an object in the detection determination area. At this time, as a technique for distinguishing and detecting a stationary object and a moving object in the detection determination area, the detection determination area is divided into a planar grid (voting box), and the number of times the object is detected is integrated (voting). ), A so-called occupancy grid method for detecting the presence of an object based on the number of times of detection (voting results) is known. In addition, when calculating the size and speed of the moving object, the size of the moving object is calculated by determining the edge of the moving object from the detection point group of the laser light irradiated to the moving object, Based on this, it is performed to track a moving body using an algorithm for identifying whether or not they are the same moving body, and calculate the speed of the moving body from the trajectory.

  Now, when detecting a moving object, not only detecting the presence of the moving object, but also identifying whether the moving object is the target moving object that is the detection target, that is, the type of the moving object The situation you want is assumed. For example, if the detection / determination area is an area where entry of a person is prohibited from the viewpoint of safety, it is necessary to specify that the moving body of the detection / determination area is a person. In this case, in the method of determining the edge of the moving object from the detection point group as described above, the position of the edge is also shifted when the moving object moves, so the size of the moving object is calculated to be larger than the actual size. It becomes impossible to detect whether the moving body is the target moving body. For this reason, when calculating the size of the moving object, it is desirable to determine the edge of the moving object from the detection point group obtained by one scan.

  On the other hand, due to the characteristics of the optical distance meter that scans the detection judgment area at every scanning angle, the end of the detection point group and the edge of the moving object do not always coincide, and an error occurs in the detection itself. In consideration of the possibility that the edge of the moving object is determined only from the detection point group obtained by one scan, there is a high possibility that the size of the moving object is erroneously calculated. Become. For this reason, for example, in Patent Document 1, a plurality of optical distance meters are provided in consideration of the possibility that the moving body is detected to be larger than the actual size when scanned multiple times with an optical distance meter. It is described that a shape correction technique for correcting erroneous detection of the center of mass of a moving body based on a group is used.

  However, the thing of patent document 1 corrects the center of mass, and there is no specific description about how the size of the moving body is corrected. In addition, in Patent Document 1, a plurality of optical distance meters are provided, each of the optical distance meters scans a plurality of times, a camera is further provided, and the detection result of the optical distance meter and the image captured by the camera are Image processing, etc., and the complexity of the apparatus is invited. Furthermore, when the detection determination area is set from the viewpoint of safety as described above, it is necessary to detect the target moving body as quickly as possible.

Special table 2011-505610 gazette

  The present invention has been made in view of the above circumstances, and an object thereof is to provide a target moving body detection method capable of specifying a target moving body in a detection determination area while reducing the time required for detection. It is in.

  According to the first aspect of the present invention, detection by the optical rangefinder when the target moving body moves in the detection determination area based on various parameters for detecting the target moving body before scanning of the detection determination area is started. The number of times is predicted, and the object is detected as a target moving body when the number of times of detection when scanning actually matches the number of times of detection predicted. In other words, the object is not identified by extracting shape information such as the edge of the object from the detection result by the optical distance meter and calculating the size and speed, but only by the number of times the object is detected. Specifies whether or not is a target moving body.

  Specifically, the maximum number of detections, which is the maximum number of times the target moving body is detected per unit time in a certain range, is calculated based on the reference width and reference speed of the target moving body set as the detection target. In addition, a virtual grid having a reference length capable of detecting the target moving body at least once in one scan is set in the detection determination area, and an object existing in the grid is the target moving body. The number of scans and the determination period necessary to determine that the Then, after calculating a predicted number of detections that is the number of times that the target moving body is predicted to be detected in the determination period based on these parameters, the detection result obtained by actually scanning the detection determination area and detecting the object Based on the above, it is determined whether or not the object is a moving body. If it is determined that the object is a moving body, a plurality of grids included in the trajectory are set as the target grid, and the total number of detections is the total number of detections in the target grid. If the number matches the predicted number of times of detection, the mobile body is detected as the target mobile body.

Thus, it is not necessary to extract the shape information of the object by detecting the target moving body based on whether or not the total number of detections (actually measured value) matches the predicted number of detections (predicted value). Therefore, even if the edge of the object cannot be correctly specified by one or several detections or the edge changes due to the movement of the object, the target moving body is specified regardless of the shape information of the object. be able to.
In addition, since it is not necessary to extract the shape information of the object and it is possible to specify whether or not the target moving body is a process with a small load of comparison between the actual measurement value and the predicted value, the detection time can be specified. It can be shortened and the target moving body can be detected quickly.

  According to the second aspect of the present invention, when the total number of detections of the target grid and the number of times of prediction detection match within a predetermined error range, the mobile body is detected as a target mobile body. For example, when the target moving body is set as a person, an error may occur in the number of detections when the hand moves or the direction of the optical distance meter changes. Therefore, by taking into consideration the occurrence of such errors and providing an error range in the number of predicted detections, the possibility of erroneous detection of the target moving body can be reduced, and detection accuracy can be improved.

  According to the third aspect of the present invention, a coefficient for weighting the number of detections is calculated according to the distance based on the distance from the optical distance meter to the grid where the object is detected, and the number of detections is weighted by the coefficient. Thus, the number of detections is adjusted according to the distance and converged to a certain value. Therefore, the target moving body can be detected by the same algorithm regardless of the distance to the object. In addition, since the coefficient is used, it is not necessary to perform processing such as calculation from the measured value of the distance, and the detection time can be shortened.

  In the invention according to claim 4, when there is a second grid adjacent to the first grid in which the number of detections is equal to or less than the upper limit number of detections and the number of detections is increased within the range of the upper limit number of detections or less, the object is Assuming the movement from one grid to the second grid, the trajectory is acquired. The grid is set with a minimum reference length that enables a target moving body having a reference width to be detected at least once in one scan. In this case, in the determination period, it is considered that another object exists in the grid (second grid) existing in the region of the reference length from the grid (first grid) where the detected object is located. Hateful. In other words, when the number of votes increases in an adjacent grid, it is highly possible that it can be determined that the object has moved to that grid. By simplifying the determination of the moving object in this way, the moving object can be determined without using a complicated algorithm for tracking the movement of the object, and the processing time, that is, the detection time can be reduced. it can.

  In the invention according to claim 5, the coefficient for performing weighting according to the positional relationship based on the relationship between the position where the optical distance meter is installed and the position where the target moving body that has entered the detection determination area is first detected is obtained. The number of times that the moving object is detected in the grid is calculated and weighted by the coefficient. Since the object enters the detection determination area from outside the detection determination area, that is, the object does not suddenly appear in front of the optical distance meter, the position and object where the optical distance meter is installed It is possible to estimate the direction of the object by grasping the relationship with the first detected position. Therefore, the same object differs by adjusting the number of detections according to the estimated direction of the object, that is, the coefficient according to the positional relationship between the position where the optical distance meter is installed and the position where the object is first detected. Even if it is detected in the direction, the number of detections can be converged to a certain value. In this case, the coefficient may be changed according to the distance or angle from the optical distance meter.

  According to the sixth aspect of the present invention, when an object is first detected in the detection determination area, the grid is set in the detection determination area by matching the position of the object with the center of the grid. For example, when the standard width of the target moving body is set as the reference length of the grid and the grid is set in advance in the detection determination area, the object may straddle two grids. In that case, it is necessary to manage the number of votes of the two grids, which may increase the processing load. Therefore, by setting a grid centered on the detected object, the object is positioned in one grid, and the processing load can be reduced. Moreover, since the processing load can be reduced, the detection time can be shortened.

  In the invention according to claim 7, when the number of detections in the determination period matches the number of detections in the immediately preceding determination period, it is determined that a stationary object exists in the grid, and the number of object detections for the grid is accumulated. And / or exclude the grid from the detection determination area. Thereby, it is possible to omit determination for the grid in the subsequent processing, and it is possible to reduce the processing load and the detection time.

The figure which shows the detection determination area in 1st Embodiment typically The figure which shows the composition of the laser distance meter typically The figure which shows the grid by the specific example 1 typically The figure which shows typically transition of the number of votes by the specific example 1 The figure which shows the detection result by the specific example 1 typically. The figure which shows the grid by the specific example 2 typically Fig. 1 schematically showing the change in the number of votes in Example 2 Fig. 2 schematically showing the change in the number of votes in Example 2 Fig. 3 schematically showing the change in the number of votes in Example 2 The figure which shows the detection result by the specific example 2 typically. The figure which shows the grid by the example 3 typically Figure 1 schematically showing the change in the number of votes in Example 3 Fig. 2 schematically showing the change in the number of votes in Example 3 The figure which shows typically the change by the distance of the number of votes in 2nd Embodiment. A diagram schematically showing the difference in the weighting of the number of votes by distance 1 Figure 2 schematically showing the difference in the number of votes by distance

Hereinafter, a target moving object detection method according to a plurality of embodiments of the present invention will be described with reference to the drawings. In addition, about the structure which is substantially common in each embodiment, the same code | symbol is attached | subjected and demonstrated.
(First embodiment)
Hereinafter, a target moving body detection method according to a first embodiment of the present invention will be described with reference to FIGS.

  As shown in FIG. 1, a laser rangefinder 1 to which the target moving body detection method of the present embodiment is applied is provided on, for example, a wall portion (not shown) of a target detection determination area 2. The laser rangefinder 1 corresponds to the optical rangefinder described in the claims. The laser rangefinder 1 detects a moving body such as a person 3 or a vehicle 4 and a stationary body 5 such as a shelf or a luggage within the detection determination area 2. As shown in FIG. 2, the laser distance meter 1 includes a control unit 10, a laser scanning unit 11, a storage unit 12, and an input / output unit 13. The control unit 10 is configured by a microcomputer having a CPU, a ROM, and a RAM (not shown), and controls the laser scanning unit 11 and the like by executing a computer program stored in the ROM.

  The storage unit 12 is configured by a non-illustrated nonvolatile memory element or the like, and functions as a storage unit that stores various data to be described later. The input / output unit 13 functions as input / output means for inputting a command signal from a higher-level control device (not shown), outputting a notification of the detection result of the moving body, as will be described later. The laser scanning unit 11 includes a laser irradiation unit 14, a mirror 15 that reflects the laser light emitted from the laser irradiation unit 14, a motor 16 that rotates the mirror 15 at a predetermined scanning angle and scanning period, and a mirror 15. And a laser light receiving unit 17 that receives the reflected light when the laser light reflected on the detection determination area 2 is reflected by the object.

  Although details will be described later, the control unit 10 sets the standard values of the lateral width and the speed of the target moving body as the detection target in the detection determination area 2 as the standard width W and the standard speed V, respectively, and within a certain range. The maximum number of detections, which is the maximum number of times the target moving body is detected per unit time, is calculated. In addition, the control unit 10 can detect a target moving body at least once in a single scan, and can be a quadrilateral virtual grid having each side having a minimum reference length (voting in the occupation grid method). (Corresponding to a box) is set in a plane within the detection determination area 2. In addition, the control unit 10 calculates the number of scans C necessary to determine that the object detected in the detection determination area 2 is a moving object, and determines that the detected object is a moving object. A necessary period is calculated as a determination period, and a predicted number of detections, which is the number of times a target mobile body is predicted to be detected in the determination period, is calculated. In addition, the control unit 10 accumulates the number of times the object is detected in the grid in the determination period as the number of detections for each grid, and if the control unit 10 determines that the object is a moving object, the control unit 10 includes the moving object. A plurality of grids are set as target grids, and the total number of detections in the target grids is calculated as the total number of detections. And the control part 10 detects the said mobile body as a target mobile body, when the total detection frequency of the target grid in a determination period corresponds to the prediction detection frequency.

Next, the operation of the laser distance meter 1 having the above configuration will be described. Each of the following processes is specifically performed by the control unit 10 of the laser distance meter 1, but for the sake of simplification of description, the laser distance meter 1 will be mainly described.
First, the laser distance meter 1 sets a target moving body that is a detection target among moving bodies that may exist in the detection determination area 2. Specifically, in this embodiment, the person 3 is set as the target mobile body among the mobile bodies existing in the detection determination area 2. Note that what kind of moving body is set as the target moving body is actually set in the laser rangefinder 1 according to the situation of the detection determination area 2 by, for example, a manager in charge of the area. .

  When the person 3 is set as the target moving body, the laser distance meter 1 sets the size of the person 3, more precisely, the standard value of the lateral width of the person 3 as the standard width W (mm). Further, the laser distance meter 1 sets a standard value of the speed when the person 3 moves (walks) as the standard speed V (mm / s). These standard width W and standard speed V are reference values for specifying that the moving body is the person 3. In particular, if the detection target is a pedestrian, for example, the standard width W = 200 mm and the standard speed V = 1000 mm / s are set. The numerical values of the standard width W and the standard speed V described above are examples when the person 3 is the target moving body, and are not limited thereto.

  Further, the laser rangefinder 1 calculates the number of scans per unit time C (cps: cycle per second) based on the scanning cycle, and the angle resolution P (mm) of the interval of the laser light at a certain distance based on the scanning angle. ). Although this angular resolution P changes according to the distance from the laser rangefinder 1, it is treated as a fixed value here for the sake of simplicity of explanation. Note that changes in the angular resolution P due to distance will be described in detail in the second embodiment.

The laser rangefinder 1 predicts the number of times that the person 3 is detected per unit time when the detection determination area 2 is scanned based on the standard width W, the standard speed V, the number of scans C, and the angle resolution P described above. The maximum number of detection times is calculated. Hereinafter, the detection of the person 3 will be referred to as “voting”, the number of times of detection will be referred to as “the number of votes”, and the maximum number of detections will be referred to as “the maximum number of votes acquired E”.
The maximum number of votes obtained E is the total number of votes obtained per unit time by an object of a certain size at a certain position, and is calculated by the following equation (1).
E = (W / P) × C (1)

  In this case, if W> P, that is, if the interval (P) of the laser beam at that position is smaller than the size of the object (standard width W), two votes can be obtained in one scan. become. Specifically, when the standard width W = 200 mm is set as described above, the angle resolution P is set as P ≦ W in order to prevent the person 3 from being detected in the gap of the laser beam. The That is, the angle resolution P is given as a maximum value of 200 mm (= standard width W). If the number of scans C is 30, for example, the maximum number of votes obtained E = (200/200) × 30 = 30 is calculated from equation (1). The number of scans C may be set arbitrarily, but here, for example, considering the possibility of cooperation with a camera or the like, it is set to 30 which is the number of frames of the NTSC standard.

  In other words, when the target moving body is the person 3 (more strictly speaking, a pedestrian), it may be an object that can obtain 30 votes per unit time in the state where it exists in the detection determination area 2. I understand. In other words, if the total number of votes as a result of removing the stationary body 5 in a certain range is a value near 30 × n (n is the number of seconds required for detection), the object is a person 3 (pedestrian). Can be specified.

  Here, the certain range corresponds to all grids in which it is determined that the person 3 has moved with the grid in the occupation grid method as one unit. This grid calculates a reference length based on a scanning angle (more precisely, angular resolution P) that enables the person 3 to be detected at least once by one scan of the laser rangefinder 1. It is set in a plane in the detection determination area 2 as a virtual quadrilateral having each side having a minimum length. Hereinafter, the length of one side of the grid is referred to as a grid size G (mm).

  In the present embodiment, the angular resolution P = 200 is set as described above. In this case, if G <P, the size of the grid is smaller than the interval between the laser beams, and there is a grid through which the laser beams do not pass, that is, a grid that cannot be detected by the person 3. That is, the grid needs to satisfy G ≧ P, and its minimum value is determined. Note that setting G> P is substantially common to the case where a plurality of laser beams pass through one grid, and the position of the moving body approaches the laser distance meter 1, and is the same as in the second embodiment. However, it can be adjusted according to the distance from the laser rangefinder 1. For this reason, taking into consideration even a reduction in the processing load, a grid size G = 200 mm through which one laser beam passes through each grid is calculated as an optimum value, as shown in FIG.

  When the grid size G is set, the laser distance meter 1 calculates the number of scans necessary to determine that the object detected in the detection determination area 2 is a moving body. Hereinafter, the number of scans necessary to determine that the object is a moving body is referred to as a scan number c. When determining the target moving body by the occupation grid method, it is important to reliably separate the target moving body from the stationary body 5 in accordance with characteristics such as the size and speed of the target moving body. Here, if the target moving body among the objects detected in the detection determination area 2 is regarded as a signal and the stationary body 5 is regarded as noise, the separation of the target moving body and the stationary body 5 is a so-called S / N ratio. There is something that leads to the idea of designing. At this time, the necessary S / N ratio is set in consideration of variations in the performance of the laser rangefinder 1.

From these facts, the number of scans c can be calculated by the following equation (2) based on the maximum number of votes obtained E, the standard speed V, the grid size G, and the S / N ratio.
c = E / (V / G) × r (2)
Where r is the S / N ratio. In the present embodiment, r is assumed to be an integer.
Specifically, as described above, since the maximum number of votes obtained E = 30, the standard speed V = 1000, and the grid size G = 200, the number of scans c = 30 / (1000/200) × r = 6 × r It becomes. That is, the number of scans c is proportional to the S / N ratio. It can be said that (V / G) in the equation (2) indicates the number of grids on which the person 3 moves per unit time.

The S / N ratio described above indicates that the signal component with respect to noise is smaller as the value is lower as in the general idea, and in this embodiment, the accuracy of detecting the target moving body is reduced. Is equal to On the other hand, since the number of scans c is proportional to the S / N ratio as described above, the greater the S / N ratio, that is, the higher the accuracy of detecting the target moving body, the more the number of scans c, that is, The time required to detect the moving object (corresponding to the determination period) becomes longer. That is, there is a trade-off relationship between the accuracy of detecting the moving object and the time required for detection. This means that the detection time can be prioritized or the detection accuracy can be prioritized depending on the situation of the detection determination area.
Hereinafter, detection of the target moving body according to the situation of the detection determination area 2 will be described with a specific example. Specific example 1 is an example of S / N ratio = 2, specific example 2 is an example of S / N ratio = 3, and specific example 3 is a case where a plurality of moving bodies exist at S / N ratio = 3. It is an example.

<Specific example 1: When S / N ratio = 2 (priority of detection time)>
When the S / N ratio = 2, the number of scans c = 12, as described above, and the determination period is 12/30 = 0.4 seconds. In this case, since the number of grids that the person 3 moves in the unit time (1 second) is V / G = 1000/200 = 5 as described above, the number of grids that the person 3 moves in the determination period is 5 ×. 0.4 = 2. Therefore, in the specific example 1, as illustrated in FIG. 3, a 3 × 3 range in the detection determination area 2 will be described as an example. Further, as shown in FIG. 3, the laser rangefinder 1 is installed in a virtual far position in the front direction of the detection determination area 2, and one laser beam is indicated by an arrow A in each grid. As shown, it is assumed that the light is irradiated almost in parallel. In addition, the predicted number of detections that the person 3 is predicted to be detected in the determination period is 30 × 0.4 = 12. Hereinafter, the number of predicted detections will be referred to as a predicted number of votes X and will be described together with the transition of the number of votes shown in FIG.

  In Specific Example 1, as shown in FIG. 4A, a person 3 (schematically shown by a cylinder in FIG. 4) and a stationary body 5 exist in the detection determination area 2, and the person 3 is in the direction of the arrow. Let's move. FIG. 4A shows an object detection result (the number of votes) at the time when the first scan is completed. Similarly, FIG. 4 (b) shows the second time, FIG. 4 (c) shows the third time, FIG. 4 (k) shows the 11th time, and FIG. 4 (l) shows the 12th time. In addition, the number of squares superimposed on each grid corresponds to the number of votes. In addition, illustration of the person 3 and the stationary body 5 is abbreviate | omitted in FIG.4 (b)-FIG.4 (l).

  The laser rangefinder 1 repeats the scanning of the detection determination area 2 every scanning cycle (1/30 second). When the person 3 enters the detection determination area 2 as shown in FIG. Voting once for that grid. In addition, since the stationary body 5 is also detected, the vote is performed once on the grid where the stationary body 5 exists. Strictly speaking, since it is not possible to determine whether it is a person 3 or a stationary body 5 at this time, the laser distance meter 1 detects an “object”. Therefore, it will be described as a person 3 and a stationary body 5. The grid in which the person 3 exists in FIG. 4A corresponds to the first grid described in the claims.

  Subsequently, the laser rangefinder 1 repeats scanning and repeats voting as shown in FIG. 4 (b) → FIG. 4 (c) → FIG. 4 (d) → FIG. 4 (e) → FIG. That is, when the laser distance meter 1 scans the detection determination area 2, the number of times that an object is detected in the grid in the determination period is accumulated for each grid as the number of detections. In this case, since the speed of the person 3 is set to 1000 mm / s, the person 3 exists in the same grid until the time of FIG. More specifically, since one scan is 1/30 second, at the time of FIG. 4F when six scans are completed, 1/30 × 6 = 1/5 ( = 0.2 seconds), and since the number of grids that the person 3 moves per unit time is 5 as described above, the person 3 is within 5 × 1/5 = 1 grid. It will stay. That is, when the person 3 having the standard width W and the standard speed V described above is assumed, the upper limit number of detections in each grid is six. Hereinafter, the upper limit detection count is referred to as the upper limit vote count.

  In the next scanning cycle, the person 3 moves to an adjacent grid as shown in FIG. The grid in which the person 3 exists in FIG.4 (g) is equivalent to the 2nd grid described in the claim. At this time, the laser distance meter 1 detects the person 3 in the second grid and votes for the second grid. On the other hand, the laser distance meter 1 further votes for the grid on which the stationary body 5 exists. At this time, the grid of the stationary body 5 is voted exceeding the upper limit vote number. In FIG. 4G and subsequent figures, voting exceeding the upper limit number of votes is indicated by hatching.

  The laser distance meter 1 repeatedly obtains the number of votes shown in FIG. 4 (l) when the determination period is completed by repeatedly performing the above-described scanning. At this time, the laser rangefinder 1 determines whether or not the detected object is a moving object based on the number of votes acquired in the determination period, and if it determines that the detected object is a moving object, acquires the trajectory. To do. Specifically, as shown in FIG. 4 (l), the laser distance meter 1 is adjacent to the first grid in which the number of votes in the determination period is equal to or less than the maximum number of votes, and the maximum number of votes in the determination period. Since there is a second grid that has increased in a range equal to or less than the number of votes, it is determined that the object is a moving body, and the movement from the first grid to the second grid is acquired as the locus of the moving body.

  In other words, since it is difficult to consider that there is another person or other moving body in the adjacent grid in the determination period, that is, within a range of 200 mm, when the number of votes increases in the adjacent grid, the object As a result, the determination of the moving object is simplified. In the case of a grid in which the number of votes exceeds the maximum number of votes, the object existing in the grid is slower than the target moving body (a stationary body or a moving body that is not the target moving body, that is, noise). Since it can be determined, it is excluded from the locus acquisition target.

Subsequently, the laser distance meter 1 sets a plurality of grids included in the trajectory as the target grid, and calculates an acquired vote number e (corresponding to the total number of detections) that is the total number of votes in the target grid. In the case of FIG. 4 (l), the laser rangefinder 1 has voted 6 times for the first grid and 6 times for the second grid, so the number of votes obtained e is 6 + 6 = 12.
The laser distance meter 1 has the number of votes acquired e (= 12) and the predicted number of votes X (= 12) as described above, so that the moving object at that position is the target moving object, that is, the person 3. Detect. Further, as shown in FIG. 5, the laser distance meter 1 has an object detected at that position that is not the person 3 because the number of votes (= 12) of the grid of the stationary body 5 exceeds the upper limit vote number. Is determined.

  As described above, the laser rangefinder 1 determines whether or not the object detected in the detection determination area 2 is a moving object, and if the moving object exists, the moving object is the target moving object (person 3 ) Or not. In the case of S / N ratio = 2, as shown in FIG. 4, when the person 3 and the stationary body 5 do not overlap the irradiation line of the laser beam, they can be detected separately, but the person 3 and the stationary body 5 can be detected separately. When the body 5 overlaps the irradiation line of the laser beam, there are some restrictions such that the stationary body 5 may enter the shadow of the person 3 and may be erroneously detected. However, when the S / N ratio = 2, since the person 3 can be detected most rapidly, for example, the detection determination area 2 is a no entry area and needs to be detected as soon as possible from the viewpoint of safety. In the case where there is, for example, by setting the S / N ratio = 2, the person 3 can be detected quickly. In this case, erroneous detection can be avoided by the operation of the laser distance meter 1 such that the stationary body 5 of less than 600 mm is not arranged in the detection determination area 2.

<Specific example 2: S / N ratio = 3 (priority of detection accuracy)>
When the S / N ratio = 3, the number of scans c = 18, and the determination period is 18/30 = 0.6 seconds. In this case, since the number of grids that the person 3 moves in the unit time (1 second) is V / G = 1000/200 = 5 as described above, the number of grids that the person 3 moves in the determination period is 5 ×. 0.6 = 3. Therefore, in the second specific example, as illustrated in FIG. 6, a 5 × 5 range in the detection determination area 2 will be described as an example. Further, the laser rangefinder 1 is assumed to be installed in a virtual distant direction in front of the detection determination area 2, and each grid is irradiated with one laser beam substantially in parallel as indicated by an arrow A. Assume that Further, the predicted number of votes X in the determination period is 30 × 0.6 = 18. Hereinafter, it will be described together with the transition of the number of votes shown in FIGS. 7 to 9 show the change in the number of votes in each control cycle, as in the first specific example. FIG. 7A shows the vote result of the first scan in the determination period, and FIG. ) Is the vote result of the 18th scan.

  When the determination period is started, the laser distance meter 1 performs the first scan, and detects an object in the detection determination area 2, that is, the person 3 and the stationary body 5 as shown in FIG. Vote for the grid you want. Subsequently, in the second scan, voting is performed as shown in FIG. 7B, and voting is repeated until the state shown in FIG. Until the time shown in FIG. 7F, the person 3 continues to exist in the same grid. That is, FIG. 7F shows the voting results when 0.2 seconds have elapsed from the start of the determination period, and the number of votes of the person 3 and the stationary body 5 is 6 times.

  Subsequently, when the person 3 moves, the laser distance meter 1 detects the person 3 in the adjacent grid and votes as shown in FIG. On the other hand, the laser distance meter 1 further votes for the grid on which the stationary body 5 exists. At this time, among the four grids occupied by the stationary body 5, the two grids on the right side of the figure are voted exceeding the upper limit vote number (= 6), and the leftmost grid in the figure moves the person 3 It is newly voted for that, and the second grid from the left end side is not voted because it entered the shadow of person 3. Votes exceeding the upper limit vote number are indicated by hatching as in the first specific example.

  Subsequently, when the person 3 further moves, the laser distance meter 1 detects the person 3 in a grid that is further adjacent as shown in FIG. 9A and votes. On the other hand, the laser distance meter 1 further votes for the grid on which the stationary body 5 exists. At this time, among the four grids occupied by the stationary body 5, the right and left grids in the figure are further voted, and the second grid from the left in the figure is voted again because the person 3 has moved, and the right side The second grid is no longer voted because it entered the shadow of person 3.

  The laser distance meter 1 repeats the above-described scanning, and obtains the number of votes shown in FIG. 9F when the determination period ends. Then, as in the first specific example, the laser distance meter 1 determines whether or not the detected object is a moving object based on the number of votes acquired in the determination period, and determines that the detected object is a moving object. The trajectory is acquired. In the case of the specific example 2, it is adjacent to the first grid in which the number of votes in the determination period is equal to or less than the upper limit vote number, and the second grid in which the number of votes increases in the range equal to or less than the upper limit vote number in the determination period. Since there is a second second grid adjacent to the second grid, it is determined that the object is a moving object, and the movement from the first grid to the second second grid is performed. Acquired as the trajectory of the moving object.

  Subsequently, the laser distance meter 1 sets a plurality of grids included in the trajectory as the target grid, and calculates an acquired vote number e (corresponding to the total number of detections) that is the total number of votes in the target grid. In the case of FIG. 9 (f), the laser distance meter 1 has voted 6 times for the first grid and 6 times for the second grid, respectively, so the number of acquired votes e is 6 + 6 + 6 = 18. Then, the laser distance meter 1 detects that the moving body is the target moving body, that is, the person 3 because the number of acquired votes e (= 18) and the predicted number of votes X (= 18) match. . Further, as shown in FIG. 10, the laser distance meter 1 determines that the detected object is not the person 3 because the number of votes of the grid of the stationary body 5 exceeds the maximum number of votes.

As described above, the laser distance meter 1 determines whether or not there is a moving body in the detection determination area 2 and, if there is a moving body, whether or not the moving body is the target moving body (person 3). doing.
In this case, it is unlikely that a plurality of people 3 simultaneously enter a 5 × 5 (1 m square) range within a determination period of 0.6 seconds. Further, as shown in FIGS. 7 to 9, since the person 3 moves even if it is temporarily in the shadow of the person 3, the person 3 and the stationary body 5 exist in the detection determination area 2. However, even when the person 3 and the stationary body 5 are overlapped on the irradiation line of the laser beam, the person 3 and the stationary body 5 can be separated. That is, in the case of S / N ratio = 3, there is no restriction as in the case of S / N ratio = 2. That is, it can be said that the state where the S / N ratio = 3 is a condition in which the person 3 can be reliably detected in the shortest time although the detection time is longer than the S / N ratio = 2.

<Specific example 3: S / N ratio = 3 and a plurality of moving objects exist in the area>
Specific examples 1 and 2 exemplify the case where the person 3 and the stationary body 5 exist. However, even if there are a plurality of moving bodies in the detection determination area 2, the movement other than the person 3 and the person 3 is performed. Can separate the body.
For example, as shown in FIG. 11, a case where a person 3 and another moving body such as a vehicle 4 enter the detection determination area 2 is assumed by taking a 5 × 5 range in the detection determination area 2 as an example. The conditions such as the standard width W, the standard speed V, the determination period, or the position of the laser distance meter 1 are the same as those in the second specific example.

  When the determination period is started, the laser distance meter 1 performs the first scan, and detects an object in the detection determination area 2, that is, the person 3 and the vehicle 4 as shown in FIG. Vote on the grid. Subsequently, in the second scan, voting is performed as shown in FIG. 12B, and voting is repeated until the state shown in FIG. Until the time of FIG. 12F, the person 3 continues to exist in the same grid. On the other hand, since the moving speed of the vehicle 4 is faster than that of the person 3, the front part of the vehicle 4 overlaps the person 3 on the laser light irradiation line at the time of FIG. 7E, and the rear part of the vehicle 4 is at the time of FIG. It overlaps with 3. Therefore, in FIGS. 7E and 7F, voting is not performed on the grid where the vehicle 4 exists (the leftmost grid).

  Subsequently, when the person 3 moves, the laser distance meter 1 votes to the adjacent grid as shown in FIG. Then, voting is repeated in the same manner as in the specific example 2, and the number of votes in FIG. 13B is obtained in the 17th scan, and the number of votes in FIG. 13C is obtained in the 18th scan. Then, the laser distance meter 1 determines whether or not the object is a moving object, and since both of the detected objects were moving objects this time, the voting result of the moving object is obtained as shown in FIG. Will get. Of the two target grids, the obtained vote number e (= 18) of the target grid on the front side of the figure (that is, the person 3 side) matches the predicted vote number X (= 18). The target mobile body, that is, the person 3 is detected. On the other hand, the obtained vote number e (= 8) of the target grid on the far side in the figure (that is, the vehicle 4 side) does not match the predicted vote number X (= 18), and the value is less than half. Since the speed is twice or more that of the person 3, it is determined that the moving body is not the person 3.

  Of all the votes (18 + 8) in the detection determination area 2, the variation (8) due to the vehicle 4 is within a removable range, and noise components (for example, wire mesh, planting, flags, curtains, etc.) It can be handled in the same manner as the detection error of the distance of the laser rangefinder 1 that occurs. Specifically, since the influence of the vehicle 4 corresponds to 288 mm with respect to the standard width W = 200 mm, it is not erroneously determined to be out of the detection target.

  As described above, the laser rangefinder 1 can reliably determine whether or not the moving body is the person 3 even when there are a plurality of moving bodies in the detection determination area 2. Can be detected as a target moving body. In addition, when the laser distance meter 1 detects the target moving body, the laser distance meter 1 outputs, for example, an indication from the input / output means. In this case, for example, the person 3 may be notified of the area where entry is prohibited by using a speaker or the like.

According to this embodiment described above, the following effects can be obtained.
Before the scanning of the detection determination area 2 is started, the number of detections by the laser rangefinder 1 when the target mobile body moves in the detection determination area 2 is predicted based on various parameters for detecting the target mobile body. When the number of times of detection (the number of acquired votes e) when actually scanned matches the number of times of detection (predicted number of votes X), the object is detected as a target moving body. In other words, the object is not identified by extracting shape information such as the edge of the object from the detection result of the laser distance meter 1 and calculating the size, speed, and the like, but depending on the number of times the object is detected. It is specified whether or not it is a target moving body.

Thereby, even in a situation where the edge of the object cannot be accurately specified or a situation where the edge changes due to the movement of the object, the target moving body can be accurately specified.
In addition, it is possible to determine whether or not the target mobile body is a target mobile body by a process that extracts shape information from the detection result of the object and that has a low load process of comparing the number of acquired votes e and the predicted number of votes X. Therefore, the detection time can be shortened, and the target moving body can be detected quickly.

(Second Embodiment)
Hereinafter, a target moving body detection method according to a second embodiment of the present invention will be described with reference to FIGS. 14 and 15. The second embodiment is different from the first embodiment in that the angular resolution is changed according to the distance. Since the method for detecting the laser distance meter and the target moving body is the same as that in the first embodiment, detailed description thereof is omitted.

  The laser distance meter 1 has a characteristic that the laser beam is irradiated radially from one point, and the number of the laser beam irradiated changes according to the distance from the object. Specifically, as the distance between the laser distance meter 1 and the object increases, the interval between the laser beams increases (the density decreases), and it becomes difficult to specify the edge of the object. In this regard, as described in the first embodiment, according to the target moving object detection method of the present application, it is possible to detect a moving object as a detection target without specifying an edge. However, since it is inevitable that the distance between the laser beams becomes wider depending on the distance, it is necessary to make some adjustment according to the distance between the laser rangefinder 1 and the object.

  Specifically, as shown in FIG. 14A, even when the same object is located near the laser distance meter 1 (the density of the laser light is high), and from the laser distance meter 1 In the case of being away (laser light density is low), the number of votes obtained by one scan (that is, the number of passing points of laser light passing through the grid) changes. In this case, the number of acquired votes e varies depending on the position of the object, and comparison with the predicted number of votes X cannot be performed correctly. Further, as shown in FIG. 14B, even if the same object is actually used, the area irradiated with the laser light may differ depending on the direction. Specifically, it is when the object is facing the front with respect to the laser rangefinder 1 and when it is facing sideways.

Therefore, in this embodiment, a coefficient for weighting the number of votes is calculated according to the position of the object, more precisely, the distance between the laser rangefinder 1 and the object, and the inequality of one vote is calculated based on the coefficient. It has been solved.
The laser rangefinder 1 sets a grid minimum value g (m), which is the minimum value of the grid at the maximum distance L, based on the angular resolution ω (rad) and the maximum distance L (m) of the detection determination area 2. . The maximum distance L is a distance from the laser rangefinder 1 to the outermost edge of the detection determination area 2. In this case, the grid minimum value g is set so as to satisfy the following (3) approximate expression.
g ≧ L × tan (ω) (3)

  That is, the grid minimum value g is set equal to or greater than the interval of the laser beams at the maximum distance L. This is because if the grid minimum value g is smaller than the interval of the laser beams at the maximum distance L, there will be a grid that cannot be voted despite the presence of the object, and the detection of the object will be inaccurate. . In the case of (3) approximate expression, the calculation of g by the trigonometric function tan () is strictly an approximate value. However, since g is sufficiently small with respect to L in an actual use situation (for example, since g = 200 mm with respect to L = 30000 mm as described later), (3) it is assumed that there is no practical problem even with the approximate expression. I use it.

Subsequently, the laser distance meter 1 calculates a coefficient Kb that cancels the difference in the number of votes per grid, that is, adjusts the weight of one vote. In a certain grid, the number of votes n per grid in one scan is the same as the number of laser beam passing points in the grid, and is calculated by the following equation (4).
n = g / (l × tan (ω)) + 1 (4)
Here, l is the distance from the laser rangefinder 1 to the object.
That is, the number of votes n increases in inverse proportion to the distance l. For this reason, the closer the object is to the laser distance meter 1, the more votes are given to one grid per position scan. This is because the number of passing points of the laser light per grid changes according to the distance from the laser distance meter 1 as described above. It can be made constant regardless. In this case, the weighting coefficient Kb can be calculated by the following equation (5).

Kb = 1 / (n / S) = S / n (5)
Here, S is the total number of votes (number of laser beam passing points) voted for one row of grids in the distance direction (laser beam irradiation direction) up to the maximum distance L, and is calculated by the following equation (6). can do.

Based on these equations, the laser rangefinder 1 votes for a value obtained by multiplying the number of times an object has been detected by a coefficient Kb corresponding to the distance. It can be said that the weight of one vote is converged to S. Specifically, the number of votes n2 when weighting is performed is calculated by the following equation (7).
n2 = n × Kb = n × (S / n) = S (7)
Thus, by adjusting the number of times (n) at which the object is detected according to the distance (by multiplying by Kb), the weighted vote number n2 becomes S. That is, even when the distance from the laser rangefinder 1 changes, the number of votes that serve as a reference for detecting the target moving body converges to S.

More specifically, an object of the same size is located at the position of the laser distance meter 1 to 30 m as shown in FIG. 15A, and the position of the laser distance meter 1 to 10 m as shown in FIG. 15B. Assuming that In this case, for example, if the scanning angle of the laser rangefinder 1 is 0.25 degrees, the angular resolution ω = 0.25 × (2π / 360). Then, assuming that the minimum grid value g is 200 mm, the number of votes (n) before weighting is calculated as follows. Note that n_30 is the number of votes when the distance is 30 m (30000 mm), and n_10 is the number of votes when the distance is 10 m (10000 mm).
n_30 = 200 / (30000 × tan (ω)) + 1 = 2.52
n_10 = 200 / (10000 × tan (ω)) + 1 = 5.58

  That is, when the distance is 10 m, the vote of one vote has a weight about 2.2 times (5.58 / 2.52) that when the distance is 30 m. If this is applied to the number of acquired votes e = 18 of the first embodiment as it is, the number of acquired votes e = 18 when the distance is 30 m, while the number of acquired votes e = about 40 when the distance is 10 m. Become. In this case, the number of acquired votes e = 40 is too large and the speed is too small with respect to the predicted number of votes X (= 18), so that it is determined that the number is not a detection target. That is, although it is a detection target, it cannot be detected.

  Therefore, the laser rangefinder 1 performs weighting according to the distance as described above, so that the same number of votes can be obtained for the same object at any distance. Accordingly, the target moving body can be detected by the same determination method (algorithm, see the first embodiment) regardless of the distance to the object. In addition, since the number of votes is adjusted by multiplying by the coefficient Kb, it is not necessary to perform processing such as calculation from the measured distance value, and the detection time can be shortened. In the above example, a coefficient that converges the vote count to S is adopted, but it is needless to say that a coefficient that converges the vote count to 1 may be adopted.

  Further, the laser distance meter 1 calculates a coefficient for performing weighting in the same manner even when the directions are different even for the same object as shown in FIG. Note that the orientation of the laser distance meter 1, that is, the positional relationship between the laser distance meter 1 and the moving body, etc. must be considered with respect to the orientation, but the illustration is omitted. For example, when the laser distance meter 1 is installed at the end of the corridor, the person moving toward the laser distance meter 1 faces the laser distance meter 1 in front. At this time, since the person enters the detection determination area 2 from outside the detection determination area 2 (that is, the person does not suddenly appear in front of the laser distance meter 1), the laser distance meter 1 is installed. The coefficient can be calculated from the relationship between the position where the person is detected and the position where the person is first detected.

  Specifically, the laser distance meter 1 can detect that the person is a person by multiplying the standard width W exemplified in the first embodiment by a coefficient so as to be a shoulder width, for example. Alternatively, when the laser distance meter 1 is installed on the wall of the hallway and the person moves so as to cross the front of the laser distance meter 1, the standard width that is a value when the person is positioned in front of the laser distance meter 1 You may make it adjust W by multiplying a coefficient according to the distance and angle from the laser rangefinder 1.

(Other embodiments)
The present invention is not limited to those exemplified in each embodiment, and can be modified or expanded as follows, for example.
The numerical values shown in each embodiment are examples and are not limited thereto.
In each embodiment, although the example which assumed the person 3 as a target moving body was shown, it is needless to say that the vehicle 4 etc. may be set as a target moving body.

  In each embodiment, when the total number of detections of the target grid (acquired vote number e) and the upper limit detection number (predicted vote number X) coincide with each other in the determination period, the target moving object is detected. When the predicted number of votes X matches within a predetermined error range, it may be detected as a target moving body. For example, in the case of the person 3, if the hand moves or the direction with respect to the optical distance meter changes, an error may occur in the number of detections. Therefore, by providing an error range for the predicted number of votes X, the target moving body can be detected in consideration of those noise components, and the detection accuracy can be improved. In this case, the error range is, for example, whether the ratio to the standard width W, the detection error of the laser rangefinder 1 or the outer shape of the target moving body changes (for example, if the vehicle 4 is used, the outer shape is considered not to change). For example).

In addition, the determination may be performed in consideration of the error with respect to the upper limit vote number that is a reference value when acquiring the trajectory of the object.
When an object is first detected in the detection determination area 2, a grid may be set in the detection determination area 2 so that the position of the object coincides with the center of the grid. For example, if the grid size G is set to the standard width W = 200 mm of the person 3 and the grid is set in advance in the detection determination area, the person 3 may exist across two grids. In that case, it is necessary to manage the number of votes of the two grids, which may increase the processing load. Therefore, by setting a grid centered on the detected object in the detection determination area 2, the object can be accommodated in one grid, and the processing load can be reduced. Moreover, since the processing load can be reduced, the detection time can be shortened.

  If the number of detections (the number of votes) in the determination period matches the number of detections in the immediately preceding determination period, it is determined that a stationary object exists in the grid, and / or voting is not performed on the grid. The grid may be excluded from the detection determination area 2. Thereby, the grid can be excluded from the determination process in the subsequent processes, the processing load can be reduced, and the detection time can be shortened.

  In each embodiment, a square grid is illustrated, but the shape of the grid is not limited to a square and may be a rectangle or a trapezoid. The shape of the grid is not limited to a quadrangle, and may be, for example, a triangle or a hexagon. In other words, the shape of the grid is arbitrary as long as there is no gap between adjacent grids, an area where an object cannot be detected is not formed, and it is possible to determine whether or not the object is a moving object. It can be set to the shape. Further, grids having different shapes such as a quadrangle and a triangle may be mixed.

  In the drawing, 1 is a laser distance meter (optical distance meter), 2 is a detection determination area, 3 is a moving body (target moving body), 4 is a moving body, and 5 is a stationary body.

Claims (6)

An object to be detected in a detection determination area set in advance is set in advance as a target moving body,
The width and speed of the reference value in the plan view of the goal mobile, respectively preset as the reference width and a reference speed,
The maximum detection count is the maximum number of times the target moving body is detected per unit time when scanning the detection determination area by an optical rangefinder having a scan angle and scanning cycle is predetermined, the Calculate in advance based on a reference width, the scanning angle and the scanning period,
A reference length capable of detecting the target moving body having the reference width at least once in one scan is calculated based on the scanning angle, and at least one side having the minimum reference length is calculated. A plurality of polygonal virtual grids that are adjacent to each other in the detection determination area,
The number of scans that is a reference value for determining whether or not the object detected in the detection determination area is a target moving body is calculated based on the maximum number of detections, the reference speed, and the reference length, Based on the calculated number of scans and the scan period, a determination period that is a period for accumulating the number of times the object is detected is calculated in advance.
Based on the maximum number of detections, a predicted number of detections that is a number of times that a target mobile body is predicted to be detected in the determination period is calculated in advance
When scanning the detection determination area with the optical distance meter, the number of times an object is detected in the grid in the determination period is accumulated for each grid as the number of detections,
It is determined whether the object is a moving body based on an increase / decrease in the number of detections in the adjacent grid in the determination period, and the number of detections is increased in a grid adjacent to the grid in which the number of detections has increased during the determination period. in the the object is determined to be a moving body, sets a plurality of grid where the object is included in the locus moves as a target grid, the sum of the number of detected grid corresponding to the determination period in the target grid Calculated as the total number of detections,
A plurality of grids included in the trajectory traveled by the object is set as a target grid, and the total number of detections in the target grid is calculated as the total number of detections.
When the total number of detections of the target grid in the determination period coincides with the predicted number of detections, an object that has moved the target grid is detected as a target moving body.   In the determination period, when the total number of detections of the target grid and the number of prediction detections coincide with each other within a predetermined error range, an object that has moved the target grid is detected as a target moving body. The target moving body detection method according to claim 1. The reference length is set to a length at which the target moving body can be detected at least once in one scan at the outermost edge of the detection determination area;
Based on the distance from the optical distance meter to the grid where the object is detected, a coefficient for performing weighting with a small weight when the distance is close to the number of detections is calculated,
3. The target moving body detection method according to claim 1, wherein the number of detections is weighted with the coefficient and accumulated. Based on the maximum number of detection times, an upper limit number of detections that is an upper limit value of the number of times that the target moving body is detected in each grid in the determination period,
When there is a second grid adjacent to the first grid in which the number of detections is equal to or less than the upper limit number of detections and the number of detections is increased within the range of the upper limit number of detections or less, the object is a moving object 4. The target moving body detection method according to claim 1, wherein a determination is made and the trajectory of the movement of the object is acquired assuming that the object has moved from the first grid to the second grid. 5. . 5. When detecting an object for the first time in the detection determination area, a grid is set in the detection determination area by matching the position of the object with the center of the grid. The target moving body detection method according to the item. If the number of detections in the determination period matches the number of detections in the immediately preceding determination period, it is determined that there is a stationary object in the grid, and the number of object detections for the grid is not accumulated, and / or The target moving object detection method according to claim 1, wherein the grid is excluded from the detection determination area.

Images