JP2009245172A - Counting device for moving object, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce a processing load by counting moving objects passing through a counting target area with a simple configuration and reducing computational complexity during counting the moving objects. <P>SOLUTION: A camera 2 is installed at the upper part of the counting target area R of moving objects T, the image of the counting target area R including moving objects T is photographed, and the photographed image is output to a counting processor 10. The counting processor 10 sets a plurality of detection point lines for the moving objects T composed of a plurality of detection points arranged in the direction crossing the passing direction of the moving objects T in the photographed image of the counting target area R along the passing direction over the entire photographed image. The counting processor 10 processes the photographed image, detects the moving objects T at each detection point, compares respective detection points, on the basis of detection results at each detection point of the plurality of detection point lines, determines presence/no presence of the moving objects T passing through the counting target area R, and counts the number of passing of the moving objects T in each passing direction, respectively. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a moving body counting apparatus and a program for photographing a counting target area of a moving body with a camera from above and counting the moving body passing through the counting target area in a predetermined direction.

  Conventionally, in order to grasp the number of people entering and leaving a building or a specific facility such as a store, for example, a moving body passing through a predetermined area such as an entrance or a passage is detected and the number of passages is counted. ing. In addition, as such a counting device, a counting target area is photographed from above by a camera, each captured image of the camera is subjected to image processing, analysis, etc., and a moving object passing through the counting target area in a predetermined direction is discriminated and counted. An apparatus is also known (see Patent Documents 1 and 2).

  However, in a conventional counting device using such a camera, a VGA (Video Graphics Array) image (640 × 480 dots) or a QVGA (Quarter VGA) image (320 × 240 dots) is used depending on the performance of the camera. In general, an image having a relatively large number of pixels is used, and image processing and analysis are performed on the entire image. Along with this, the acquired image also needs to be calculated and analyzed by a complicated algorithm, and complicated and troublesome processing and calculation are required for counting moving objects. As a result, in the conventional counting device, the amount of calculation required is very large and the processing load tends to increase. To cope with this, an arithmetic processing device such as a high-performance CPU (Central Processing Unit) And a large-capacity memory are required, which increases the cost of the entire apparatus.

JP 2006-285409 A JP-A-63-85890

  The present invention has been made in view of the above-described conventional problems, and an object of the present invention is to make it possible to count moving objects passing through the counting target area with a simple configuration, and to reduce the amount of calculation when counting moving objects. Thus, the processing load is reduced and the cost of the counting device is reduced.

The present invention is a moving body counting device that counts a moving body that is photographed by a camera from above and that passes through the counting target area, and is arranged in a direction that intersects the passing direction of the moving body on a captured image of the counting target area A detecting unit for setting a plurality of detection point sequences of the moving body composed of a plurality of detection points along the passing direction, and detecting the moving body for each detection point of each detection point sequence from the captured image of the counting target region. Detection means, passage determination means for determining the presence or absence of a moving body that passes the counting target area based on the detection result of the moving body at each detection point of the plurality of detection point sequences, and passing the counting target area in the passing direction And a counting means for counting the number of moving bodies to be operated.
Further, the present invention is a program for causing a computer to function as each unit of the moving body counting device according to any one of claims 1 to 8.

  According to the present invention, it is possible to count moving bodies that pass through the counting target region with a simple configuration, reduce the calculation amount when counting moving bodies, reduce the processing load, and reduce the cost of the counting device. it can.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
The moving body counting device (hereinafter referred to as a counting device) of the present embodiment is a device that counts the number of moving bodies that pass through the counting target region by photographing with a camera from above, for example, a building or facility such as a store. It is installed in the vicinity of a door, a passage, etc., and used for counting moving objects such as people passing through the area in a predetermined direction. Below, the case where the moving body which moves through a door is counted inside the door which is the doorway to a building is taken and demonstrated as an example. In this case, the counting device uses a camera that captures a low-pixel image, and counts the number of moving bodies that move in and out through the door and enter and leave the building for each passing direction.

FIG. 1 is a schematic diagram showing a schematic configuration of the counting device of the present embodiment, and the installation state of the building 90 inside the door 91 (for example, a sliding automatic door) (counting target region R) is also from the side. This is shown schematically.
As shown in the figure, the counting device 1 includes a counting processing device 10 that also serves as a control device, and a camera 2 connected to the counting processing device 10. (In this case, a person) is detected, and the number of passages (the number for each arrow direction in the figure) is counted.

  The camera 2 is a photographing means (camera module) of the counting target region R set in a predetermined range in the building 90, and is installed downward above the passage of the building 90, and the entire counting target region R is The image is taken from above with the moving body T moving above. Here, the counting target region R is set so as to form a substantially rectangular shape in the vicinity of the door 91 so as to cross the passage of the building 90, and the camera 2 is positioned above the center of the counting target region R and is light The shaft 90 is attached to the ceiling of the building 90 with the axis directed substantially downward in the vertical direction. The camera 2 includes a commercially available digital camera, a digital video camera, a CMOS (Complementary Metal Oxide Semiconductor) image sensor, or the like, and is controlled by the counting processing device 10 so that images in the counting target region R are continuously or Photographs are taken at predetermined intervals and timing, and the photographed images are output to the counting processing device 10.

  The counting apparatus 1 uses a camera 2 that captures a color image of relatively low pixels (40 × 30 dots in this case), thereby capturing the counting target region R and the moving body T to obtain a color. Output an image. At that time, the camera 2 converts the image formed on the imaging element into an electrical signal and performs A / D conversion, performs various correction processes by the built-in image signal processing circuit, and converts the image data of the captured image. The data is output to the counting processing device 10.

  The count processing device 10 includes a central processing unit (CPU) 11, a ROM (Read Only Memory) 12 that stores programs for causing the CPU 11 to execute various control and counting procedures, and CPU processing data. And a RAM (Random Access Memory) 13 for temporarily storing programs and programs. Further, the counting processing device 10 includes a setting input unit including a setting dip switch and the like, and an input / output unit (not shown) having an input / output port to which the camera 2 and the like are connected, and each connected Various data and control signals are transmitted to and received from the device. As a result, the counting processing device 10 controls the camera 2 to shoot based on a predetermined program, preset conditions, and the like, and performs image processing and analysis on the acquired captured image. Each process for counting is executed.

FIG. 2 is a functional block diagram showing a schematic configuration of the counting processing device 10.
As shown in the figure, the counting processing device 10 includes a control unit 14 that controls the entire device and counting processing, and a storage unit 15 that stores various data such as data acquired from images taken by the camera 2 and data being processed. In addition, a setting unit 16, a counting unit 17, and an image processing unit (analysis unit) 20 are provided. The counting processing device 10 includes the setting input unit 18 and the input / output unit 19 to which the output destination of the camera 2 and the counting result is connected. These units (means) 14 to 20 are connected via the bus K or the like. Connected to each other.

  The setting input unit 18 is used for setting the installation conditions of the camera 2, the processing conditions of the counting processing device 10, etc., and outputs the contents input from the user to the control unit 14 or stores them in the storage unit 15. From the setting input unit 18, for example, the installation height of the camera 2 and shooting conditions are input and set, and based on the information, the control unit 14 sets the setting contents by the setting unit 16 to be described later based on preset conditions. Various conditions relating to the detection of the moving body T and the counting process are set.

  The setting unit 16 is a means for setting a specific detection point (extraction point) for detecting the moving object T with respect to the captured image acquired from the camera 2 via the input / output unit 19. A plurality of pixel positions to be used for processing are set according to each condition. The count processing device 10 extracts the detection point data of the moving body T set by the setting unit 16 from each captured image, and executes the detection of the moving body T and the counting process using the data.

  FIG. 3 is a schematic diagram for explaining setting of detection points by the setting unit 16, and FIG. 3A shows an example of a captured image G of the counting target region R by the camera 2. In FIG. 3A, the vertical direction is the passing direction of the moving body T (see FIG. 1), and along that direction (see the two arrows on the right side of FIG. 3A), the moving body T captures the captured image G (count target area). Move in R).

  As shown in FIG. 3A, the setting unit 16 includes a plurality of (five in the left and right directions in the figure) directions in which the captured image G of the counting target region R acquired from the camera 2 intersects the passing direction of the moving body T. The detection points P (shown by rectangles in the figure) are arranged apart from each other, and a detection point sequence of the moving body T including the plurality of detection points P is set. In addition, a plurality of detection point rows are set apart from each other along the passing direction (vertical direction in the drawing) of the moving body T (rows A, B, and C shown in the drawing), and the captured image G The detection points P are set substantially evenly in a state of being separated from each other.

  At the time of this setting, in the present embodiment, the plurality of detection points P are arranged at substantially equal intervals along the passing direction of the moving body T and the direction orthogonal thereto. Further, each detection point P has a size such that only one moving body T is detected, for example, 1 to several pixels in length and width (here, 1 to 2 pixels in this case), and the size of the captured image G Depending on the number of dots and the like, they are arranged at a predetermined dot interval (here, an interval of about 6 dots in length and width) that does not cause detection of the moving body T. At this time, in this counting device 1, in each detection point row, adjacent detection points P are arranged according to the shoulder width of an adult, and one adult detects at one detection point P or two adjacent detection points P. To be. In addition, the setting unit 16 sequentially arranges first, second, and third detection point rows arranged along the passing direction of the moving body T (for example, when the passing direction is a downward direction in FIG. Column (corresponding to column C) is set at least (here, only these three columns are set). The counting processing device 10 determines the moving (passing) direction of the moving body T based on the detection result of the moving body T at each detection point P in each row, its change, and the like, and the presence or absence of the moving body T passing through. Judgment etc.

In this way, the count processing device 10 sets the detection point sequence of three columns (A column, B column, C column) each including five detection points P to the captured image G by the setting unit 16, Data of the detection point P is extracted (see FIG. 3B) and processed by the image processing unit 20.
Note that FIG. 3B schematically shows a state (image) in which only the detection point P is extracted. In the following, mainly using the same block diagram, the counting process of the moving object T, etc. Will be described. Further, when particularly distinguishing each detection point column, it is represented only as column A, column B, and column C (see the left side in FIG. 3B), and these five detection points P are numbers 1 to 5 in order from the left. (See the upper side of FIG. 3B) for distinction. For example, in the A column, from the left, the A column 1, the A column 2, the A column 3, the A column 4, the A column 5 (the same applies to the B column and the C column). To express. Further, in the following description, the direction from the A row to the C row is the down direction, and the opposite direction is the up direction.

  The image processing unit 20 (see FIG. 2) is means for acquiring image data of each detection point P from the captured image G and performing image processing and analysis processing. The detection unit 21, the number determination unit 22, and the comparison The passage determination unit 23 includes a unit 24 and a moving body determination unit 25.

  The detection unit 21 is a means for detecting the moving body T for each detection point P in each detection point sequence from the captured image G in the counting target region R. For example, each detection point of the frame image continuously captured by the camera 2 is detected. The Ps are compared, the difference between the frames is calculated, and the image change is extracted for each detection point P. From this result, the detecting unit 21 compares each difference amount with a predetermined threshold value to determine whether or not there is a moving body T at the position of each detection point P, and each detection point P in the detection point sequence. To detect the moving object T.

  Based on the detection results at the plurality of detection points P of each detection point sequence, the number determination unit 22 uses the detection point sequence based on the number of detection points P that have detected the moving object T, their positional relationship, and the like. The number of detected moving objects is determined. At this time, the number conversion table set in advance is used here, and the number of the moving bodies T detected by associating the table with the detection point P where the moving body T is detected is determined.

FIG. 4 is a schematic diagram showing a part of this number conversion table. In each stage, detection results at detection points P (corresponding to detection points 1 to 5 in FIG. 3B) and the results are shown in each stage. The number of the moving bodies T corresponding to is shown. In FIG. 4, the detection point P where the moving body T is detected is 1 and the non-detection detection point P is 0.
FIG. 5 is a schematic diagram for explaining the determination by the number determination unit 22. The captured image G of the count target region R similar to that in FIG. 3A and the moving body T at each detection point P in the captured image G are shown. The four detection examples are shown side by side. Further, in FIG. 5, the detection result of the moving body T is shown by a block diagram similar to FIG. 3B, and 1 is written in the detection point P in which the moving body T is detected. On the other hand, in the captured image G, each moving body T is schematically shown as an ellipse, and the moving direction is also indicated by an arrow.

  In the present embodiment, as shown in FIG. 4, with respect to the state where the moving body T is not detected (see FIG. 5A), the moving body T is moved at one or two adjacent detection points P in each detection point row. When it is detected (see FIGS. 5B and 5D), the number of moving bodies T is set to 1. Further, when they appear at two and three positions with one or more detection points P interposed therebetween (see FIG. 5C), the numbers of the moving bodies T are determined to be 2 and 3, respectively. On the other hand, when the moving body T is detected at three or four consecutive detection points P, the number of the moving bodies T is determined as 2 when detected at all 5 and 3 when detected at all five, and these are combined, The number of moving bodies T detected in the detection point sequence is determined. The image processing unit 20 determines the number of moving bodies T by converting the detection results at a plurality of detection points P based on the number conversion table set in this way by the number determination unit 22.

  The passage determination unit 23 (see FIG. 2) compares or acquires, for example, all or some of the detection point sequences based on the detection result of the moving body T at each detection point P of the plurality of detection point sequences. The presence or absence of the moving body T that passes through the counting target region R is determined from the change in the detection position between the captured images G or the like. At the time of this determination, the passage determination unit 23 acquires the detection result of the moving body T with respect to the captured image G of the current frame or the previous frame, and the comparison unit 24 according to a preset condition, The positions (1 to 5 in FIG. 3B) of the detection points P that detect the moving body T between the detection point sequences are compared. In addition, the passage determination unit 23 reduces the count target region R to the above-described down state by, for example, comparing the comparison results of predetermined detection point sequences with each other by the moving body determination unit 25 based on the comparison result of the positions. The moving body T that moves in either the direction or the passing direction is discriminated. Based on the determination result, the passage determination unit 23 determines the presence or absence of the moving body T that passes through the counting target region R in both directions.

  At that time, the passage determining unit 23 moves the respective moving directions in the respective passing directions by the moving body discriminating unit 25 based on the change in the position along each passing direction of the detection point P that has detected the moving body T between the detection point sequences. The body T is discriminated, and it is determined whether or not the moving body T has passed the counting target region R. However, here, in order to ascertain whether or not the moving object T that has been skewed or meandered in the counting target region R has passed and has been reliably determined, the detection points P at corresponding positions between the respective detection point sequences. In addition, the detection point P on both sides of the adjacent one is also compared to determine the moving body T.

  Further, in the counting device 1, the passage determination unit 23 includes the first detection point row located on the upstream side in the passage direction of the moving body T and the third detection point row located on the downstream side in the passage direction (each A row). Or the C column), the mobile body T newly detected in each detection point sequence is grasped based on the detection result of the mobile body T for each detection point P. As a result, the position and number of the moving body T entering the counting target region R and the position and number of the moving body T moving out of the counting target region R are determined. Furthermore, the passage determination unit 23 compares the position and number of the detection points P that detect the moving body T in the third detection point row and the second detection point row (B row) located in the middle of the passage direction. Then, the moving body T moving in each passing direction is discriminated, and these discrimination results are compared. When the passage determination unit 23 detects the moving body T that goes out of the counting target region R by the determination in the third detection point sequence, the passage determination unit 23 executes a determination process and compares the above determination results, and the comparison result Based on the above, the presence or absence of the moving body T passing through the counting target region R and the passing direction are determined.

  In this way, the counting processing device 10 determines the passage of the moving body T and the like, and based on the determination result, the counting unit 17 causes the number of moving bodies T that pass through the counting target region R in each passing direction ( The number of passes) is counted and counted sequentially each time it passes. At that time, the counting unit 17 counts the moving bodies T that pass through the counting target region R in opposite directions, and counts the moving bodies T for each passing direction. In addition, as will be described later, the counting processing device 10 monitors the detection result for each detection point P by the control unit 14, and continuously detects all the detection points with a predetermined number (frames) of the captured images G. On the condition that the moving body T is not detected in P, the passage determination unit 23 clears the fluctuation data regarding the determination in use. Thereby, the passage determination unit 23 is reset, the data and the like are returned to the initial state at the start of the determination process, the determination is newly started, and the determination of the presence / absence of the moving mobile body T is restarted.

  Here, in the present embodiment, data relating to the detection result of each detection point sequence is processed in 1 byte (8 bits), and data of 5 detection points P (see FIG. 3) is converted into each bit in 1 byte. Data is processed correspondingly. In addition, the data of the three detection point columns (A column, B column, C column) is held in a register and compared, and the counting process including the process of the passage determination unit 23 is executed. .

FIG. 6 is a schematic diagram for explaining the data processing, and shows an example of data in three columns (3 × 5) corresponding to FIG. 3B.
In this embodiment, as shown in the figure, 5 bits (portion indicated by a solid line in the figure) of 8 bits are used for each of the A column, the B column, and the C column, and the other 3 bits (portion indicated by a broken line in the drawing). ) Is fixed at 0 to process data. In this way, the data of each detection point P in the detection point sequence is represented by 1 bit, and the bit operation is performed by setting the bit corresponding to the detection point P that detected the moving body T to 1 and the other bits to 0. . Hereinafter, a specific example of processing by the above-described image processing unit 20 and the like in which data (0 or 1) relating to detection of the moving body T at the detection point P is assigned to each bit in this way will be described in detail. In each figure, only the 5-bit portion used is shown, and the other 3 bits are omitted. Also, in each figure, the bit (the bit at which data is 1) in which the mobile object T is detected is hatched with a diagonal line to distinguish it from other blank bits (the bit at which data becomes 0).

FIG. 7 is a schematic diagram for explaining processing when the moving object T is detected in the detection point sequence. FIG. 7A is from the captured image G of the current frame, and FIG. 7B is from the captured image G of the previous frame. The detection result of the acquired moving body T is shown.
As shown in the figure, the image processing unit 20 (passage determination unit 23) monitors a current predetermined detection point sequence (here, A column and C column) and relates to detection of the moving object T at each detection point sequence. From the data (hereinafter referred to as detection data) and the corresponding detection data of the same detection point sequence in the past, in order to grasp the change of the detection point P where the moving body T is detected in the detection point sequence, the detection position, etc. Is calculated and stored.

  In this case, first, an exclusive OR (hereinafter referred to as XOR) is calculated between the present and previous (previous) detection data (hereinafter referred to as current data and previous data, respectively) in the detection point sequence. . Subsequently, a logical product (hereinafter referred to as AND) of the calculated data and the current data is calculated ((current data XOR previous data) AND current data), and compared to the previous data, the detection point sequence of the current data Then, the data (hereinafter referred to as change data) relating to the detection point P that has been changed by newly detecting the moving body T is calculated. Next, in (in) data (hereinafter referred to as in-data) relating to the moving body T that has entered each detection point sequence is calculated and stored for each detection point sequence from the change data. This in-data is the change data calculated when there is no in-data when calculating the change data. When there is in-data in the past, the in-data is the logical sum (hereinafter referred to as OR) of the calculated change data. Calculated by calculation (in-data OR change data).

  Specifically, the change data and in data of the A column (hereinafter referred to as A column change data and A column in data, respectively) are (10100) from the current data (10100) and the previous data (00000). On the other hand, change data and in data of the C column (hereinafter referred to as C column change data and C column in data, respectively) are (00100) from the current data (00100) and the previous data (00000).

FIG. 8 is a schematic diagram for explaining the comparison process between the detection point rows and the in-data detection point P. When the moving body T moves from the A row toward the C row in the down direction, FIG. An example is shown.
Here, the image processing unit 20 detects the moving object T at the detection point P (C column 3) of the C column (see FIG. 8A) and changes the C object. The change data (changed detection point P) and the past B column (previous data) are compared to determine the presence or absence of the moving body T that has moved from the B column to the C column. At this time, when the moving body T is detected at the detection point P (B row 3) (see FIG. 8B) at the same position in the B row (previous data), there is a moving body T that has moved from the B row to the C row. Is determined. In addition, as described above, when the moving object T is detected at the detection point P (B column 2 or B column 4) (see FIG. 8C and FIG. 8D) adjacent to the column B (previous data). , It is determined that the moving body T has moved from the B column to the C column in a skewed manner. On the other hand, when the moving body T is detected at the detection point P (for example, B row 1) (see FIG. 8E) at a position apart from these, it is determined that there is no moving body T that has moved from the B row to the C row. .

  Thereafter, when the image processing unit 20 determines that there is a moving body T that has moved from the B row to the C row, the moving data related to the moving body T (the same data as FIG. 8A) and the A row in data (not shown). ) In the same manner, and when the moving body T is detected at the corresponding detection point P (A row 2, A row 3, A row 4), the moving body T that has moved from the A row to the C row is Judge that there is. In this way, the presence or absence of the moving body T passing through the counting target region R is determined, and the number of moving bodies T passing in the down direction is determined from the movement data and the number conversion table (see FIG. 4). Further, this number is compared with the number of moving bodies T that have entered the counting target region R determined from the A-column in data and the number conversion table, and when the number is less than that, the number is counted as a passing number. As a detection error, the number determined from the in-column A data is counted as a passing number. Further, the image processing unit 20 turns on a flag (hereinafter referred to as a count flag) indicating whether the count process is being performed (the number of passages has already been counted) or the A column in-data. Then, a process of clearing the detection point P at the position corresponding to the counted moving body T as the moving body T has already passed is also executed.

  As described above, the counting device 1 constitutes each means for executing each processing related to counting by the counting processing device 10 and is controlled and processed by the counting processing device 10 based on a predetermined program, conditions, etc. From the captured image G of the counting target region R acquired sequentially from the above, a counting process of the number of passes of the moving body T described later is executed. Further, when the moving body T moves in the opposite direction, that is, in the upward direction from the C row toward the A row (see FIG. 3B), the counting processing device 10 performs the above arithmetic processing in the A row and the C row. The data relating to the above are exchanged in the same manner, the number of passages is sequentially counted, and the number of moving bodies T passing in both directions is counted.

Next, the counting process of the moving body T by the counting device 1 will be described in detail.
FIG. 9 is a flowchart showing a part of the counting process procedure by the counting device 1, and shows the flow of the counting process of the moving body T passing in the down direction from the A column toward the C column. On the other hand, FIG. 10 is a schematic diagram showing a specific example of the counting process in the same manner as in FIGS. 6 to 8, and each data from the photographed images G (frame images) taken sequentially is directed from the top to the bottom. Are shown in order. In addition, in FIG. 10, the detection data of columns A to C, column A change data (upper) and column A in data (lower), and column C change data (upper) and column C in data (lower) are sequentially displayed from the left. Is shown for each captured image G.

  In the present embodiment, in the present embodiment, as shown in FIG. 9, in order to monitor the moving body T that enters the counting target region R from the detection data of the A column, the above-described A column change is performed by the counting processing device 10. The data and the A-column in data are sequentially calculated by the above-described arithmetic expressions and stored in a register or the like (S101). In the example shown in FIG. 10, the moving object T is first detected in the A column 3 of the second frame (FIG. 10B), and then the moving object T is newly detected in the A column 2 of the fourth frame (FIG. 10D). Thus, the seventh frame (FIG. 10G) is maintained, and the A column change data and the A column in data change accordingly (see FIGS. 10A to 10G).

  Further, the counting device 1 determines whether or not the moving object T is detected in all of the A column, the B column, and the C column (S102), and when the moving object T is detected (S102, NO), the B column Whether or not there is a change in the detected data in column C is determined by comparing the current data with the previous data of the previous frame (S103). As a result, when there is a change in at least one column (current data ≠ previous data) (S103, YES), it is determined whether there is an increase in the C column from the C column change data sequentially calculated in conjunction with the above processing. A determination is made (S104). In the example shown in FIG. 10, the B column detects the moving body T in the seventh frame (FIG. 10G), the B column 2 and the B column 3 change, and the C column is the C frame in the tenth frame (FIG. 10J). The moving body T is detected in the column 2, and the C column change data and the C column in data at the position are also changed to cause an increase in the C column.

  Thus, when the C column has changed (S104, YES), the counting device 1 determines whether or not there is a movement of the moving body T from the B column to the C column (S105). For this determination, as described above (see FIG. 8), the current C column change data is compared with the past B column detection data (B column previous data). When either the detection position of T or the position on both sides thereof matches the change position of the C column change data, it is determined that the moving body T has moved. Specifically, the logical sum (OR) of the B-column previous data and the two previous data obtained by shifting the detection position of the moving body T one by one is calculated, and the result and the C-column change data are calculated. AND (C column change data AND (B column previous data OR B column previous data >> 1 OR B column previous data << 1)). When the calculation result (hereinafter referred to as movement data) is not 0, it is determined that there is movement (S105, YES). In the example shown in FIG. 10, the C column change data of the 10th frame (FIG. 10J) is compared with the detection data of the B column of the 9th frame (FIG. 10I), and the calculated position of the B column (B column 1, B Since the change position of the C column (C column 2) is included in the column 2, the B column 3, and the B column 4), the moving data is not 0, and it is determined that the moving body T is present.

  Subsequently, the movement data is compared with the current A column in data in the same manner (see FIG. 8) and operated (movement data AND (A column in data OR A column in data >> 1 OR A column in data << 1)), and it is determined whether or not there is a detection corresponding to the A column in-data (S106). As a result, when there is a corresponding detection (S106, YES), it is determined that there is a moving body T that moves through the counting target region R in the down direction, and the counting process is started (S107). In the example shown in FIG. 10, the detection position of the movement data (C column 2) and the calculation position (A column 1, A column 2, A column 3, A column 4) of the A column in data (FIG. 10J) are compared. Since the position corresponding to the A-line in data is detected, it is determined that there is a moving moving body T and the counting process is started.

FIG. 11 is a flowchart showing the procedure of the counting process performed by the counting device 1.
In the count process, as shown in the figure, first, it is confirmed whether or not the count flag is off (OFF) (S201). When the count is off for the first time (S201, YES), The number conversion table (see FIG. 4) is searched to determine the number of moving bodies T scheduled to pass (hereinafter referred to as number A) (S202). Next, the number conversion table is similarly searched from the movement data to determine the number of moving bodies T (hereinafter referred to as number B) that pass through (S203), and these are compared (S204). As a result, when the number B is larger than the number A (number B> number A) (S204, YES), the number A determined from the in-column A data is counted without adopting the number B determined from the moving data. (S205). On the other hand, when the number B is equal to or less than the number A (number B ≦ number A) (S204, NO), the number B determined from the movement data is counted (S206). In the example shown in FIG. 10, the number B determined from the movement data (see FIG. 10J) is 1, whereas the number A determined from the A column in data is also 1, and 1 of the number B is counted.

  Next, the counting device 1 turns on the count flag (S207), and counts the number of moving bodies T (cumulative) (hereinafter referred to as count number) and the movement data (hereinafter referred to as count data) subjected to count processing at that time. Is saved (S208). Further, the data in the position corresponding to the counted moving body T in the A-line in data is cleared to 0 (S209), and then the process proceeds to the counting process of the moving body T moving in the up direction (S210). Here, this clearing process is performed by calculating the logical product (AND) of the complement (NOT) of the moving data and the A column in data (A column in data AND (NOT moving data)). It will be used later as a long A-in data. In the example shown in FIG. 10, A column 2 at the position corresponding to the movement data is cleared from the A column in data before clearing (FIG. 10J), and new A column in data in which only A column 3 is the detection position. (FIG. 10K) is calculated.

  On the other hand, when the count flag is on and counting is continued (S201, NO), first, the number conversion table is searched in the same manner as described above, and the number A is determined from the A column in-data (S211). . In addition, new count data is calculated from the logical sum of the current movement data and the count data stored at the previous count (S208) (movement data OR count data). (S212). Next, the number C is compared with the current count number (S213). When the number C is equal to or less than the count number (number C ≦ count number) (S213, NO), the process proceeds to S208 without performing a new count. On the other hand, when the number C is larger than the count number (number C> count number) (S213, YES), a difference is calculated by subtracting the count number from the number C (S214), and the difference is compared with the number A (S215). ). As a result, when the difference is larger than the number A (difference> number A) (S215, YES), the number A determined from the A column in-data without using the difference is counted (S216). Below (difference ≦ number A) (S215, NO), the difference is newly counted (S217). Thereafter, the process proceeds to S208, and the respective steps are executed in the same manner as described above to proceed to the counting process in the up direction (S208 to S210).

FIG. 12 is a schematic diagram showing an example of the counting process when the count flag is turned on as shown in FIG.
Here, as shown in the figure, first, the moving body T is detected first in the A column 3 and A column 4 of the first frame (FIG. 12A), and then in the A column 1 of the second frame (FIG. 12B). The moving body T is newly detected at a distant position, and all of them disappear at the sixth frame (FIG. 12F). Meanwhile, the A column change data and the A column in data are sequentially calculated in the same manner as described above, and change in accordance with the change in the detection position of the moving body T (see FIGS. 12A to 12F). Further, in the B row, as the moving body T entering the A row moves in the down direction, after the moving body T is detected for the first time in the third frame (FIG. 12C) (B row 4), the detection position is It gradually increases (see FIGS. 12C to 12G).

  On the other hand, in the C column, the moving body T is detected for the first time in the eighth frame (FIG. 12H) (C column 3), and the C column change data and C column in data at the same position also change. In response to this change, the counting device 1 performs the counting process of the moving bodies T in the same manner as described above, and determines the number of moving bodies T. Here, the B column (FIG. 12G) of the previous frame has a position detection (B column 3) corresponding to the detection position C column 3 in the C column, and also corresponds to the A column in-data (FIG. 12H). Since there is a detection (A column 3), the counting process when the count flag is off is started (S202). At this time, the number A determined from the A column in data is 2, whereas the number B determined from the moving data in the C column is 1. Therefore, a smaller number B of 1 is counted and counted. The flag is also turned on. Further, the data in the position (A column 3) corresponding to the counted moving body T in the A column in data is not cleared (FIG. 12I).

  Subsequently, when the new moving body T is detected in the C row in the 11th frame (FIG. 12K) (C row 1), the counting process of the moving body T is performed in the same manner, and the count flag is on. Count processing starts (S211). At this time, here, the number C is 2 from the new count data (C column 1, C column 3) obtained by ORing the current movement data (C column 1) and the previously stored count data (C column 3). It is determined. This number C is larger than 1 of the already counted count number and the difference 1 is smaller than 2 of the number A determined from the A column in-data, so that the difference 1 is newly counted and the count number Becomes 2. Thereafter, the same processing as described above is executed, such as clearing the A-column in data and saving the count number and new count data.

  In the present embodiment, when there is no detection of the moving object T in all of the columns A to C (see FIG. 9) (S102, YES), the counting is performed when this non-detection state continues for three frames. The flag is cleared and turned off, and predetermined data such as stored count data is cleared (S108), and the process proceeds to the counting process in the up direction (S109). Thereby, the conditions up to that time are cleared and returned to the initial state, and when the moving object T is next detected, the detection process is newly resumed. If it is determined NO in each of the above steps (S103, S104, S105, S106), the process proceeds to the counting process in the up direction (S109), and based on the same detection data in each detection point sequence The moving body T passing in the direction is determined and counted.

  However, when it is determined that there is no increase in the C column (S104, NO), it is determined whether or not the detection position is decreased in the B column and the C column, compared to when the moving object T is counted before that. (S110). This determination is based on the count data saved at the time of the count processing (S208) and the detection data (B column current data, C column current data) of the current B column and C column. A logical product (AND) with the logical sum (OR) of the column current data is calculated (count data AND (B column current data OR C column current data)) and determined. As a result, if the calculation result does not match the count data (calculation result ≠ count data), it is determined that a decrease has occurred in the B and C columns (S110, YES), and the count flag, count data, etc. are set in the same manner as described above. After clearing (S111), the process proceeds to the counting process in the up direction (S109). On the other hand, when there is no decrease (calculation result = count data) (S110, NO), the process proceeds to the counting process in the up direction (S109), moves from the C column toward the A column (see FIG. 3B) and passes. The counting process of the moving body T is performed.

FIGS. 13 and 14 are flowcharts showing a part of the counting process procedure in the up direction by the counting device 1, FIG. 13 shows the procedure before the counting process, and FIG. 14 shows the procedure of the counting process. .
The counting process in the up direction is sequentially executed by the same process or calculation by replacing each data related to the A column and the C column with respect to the down direction process (see FIGS. 9 and 11) described above, and then again. Proceed to processing in the down direction. Accordingly, in FIGS. 13 and 14, steps corresponding to the steps in FIGS. 9 and 11 are denoted by “′” at the end thereof, for example, S 101 ′, S 201 ′, etc., and description thereof is omitted. .

  As described above, in the counting device 1 according to the present embodiment, a plurality of detection point sequences (detection points P) are set in the captured image G of the counting target region R, and the moving object T is detected by them, and the bit is detected. The number of passages is counted by determining the presence or absence of passage of the moving body T by calculation. Therefore, even the captured image G with a small number of pixels can be accurately counted by determining the passage of the moving body T, the camera 2 that captures a low-pixel image can be used, and the counting process can be performed easily. In addition, since it is not necessary to perform image processing on the entire captured image G, and only a part of the captured image G is used for counting, the algorithm is relatively simple, and the amount of calculation required for counting the moving object T is reduced. Thus, the processing load can be reduced. Accordingly, a low-priced CPU with a low processing speed can be used, and the required memory capacity can be reduced, so that the cost of the entire apparatus can be reduced.

  Further, the position of the detection point P at which the moving body T is detected is compared between the detection point sequences, and the moving body T is determined based on the presence or absence of detection at the detection point P at the corresponding position. The movement direction of the body T and the passage of the counting target region R can be accurately and reliably discriminated, and the number of passages can be accurately counted. Furthermore, in this counting device 1, the detection points P that detect the moving body T between the predetermined detection point sequences are compared with each other, and from the comparison results with the detection points P at the corresponding positions and the detection points P on both sides, etc. The change of the position along the passing direction of the moving body T is grasped. Since the moving body T that moves in the passing direction is discriminated based on the change in the position between the detection point rows obtained in this way, the discrimination accuracy can be effectively increased by a relatively simple process. In addition, since the number of moving bodies T detected by the detection point sequence is determined and acquired based on the detection results of the detection point sequence, the number of passes of the moving body T can be accurately counted. At this time, since the number of the moving bodies T is determined using the number conversion table (see FIG. 4), the determination process can be easily performed.

  Therefore, according to the present embodiment, it is possible to accurately count the moving body T passing through the counting target region R with a simple configuration, reduce the calculation amount when counting the moving body T, and reduce the processing load. The cost of the apparatus 1 can be reduced. In addition, since the counting device 1 counts the moving bodies T that pass through the counting target region R in opposite directions, the number of passages in both directions can be grasped, and the convenience of the counting device 1 can be ensured. At this time, since the counting in both directions is performed by the same calculation processing using the data detected and calculated in the same manner, the increase in the calculation amount and the data amount is suppressed, and the processing is relatively simple. Can be implemented. In addition, the counting device 1 clears the count flag and the like when the moving body T is not continuously detected in the predetermined number of captured images G, returns to the initial state, and determines whether or not the moving body T has passed. In order to resume, unnecessary data and the like before that can be cleared to prevent them from affecting the counting process. Along with this, it is possible to cope with the occurrence of a detection error of the moving body T, the accumulation of unnecessary data during the counting process, and the like, and it is possible to suppress a decrease in counting accuracy.

  Here, in the present embodiment, at least three detection point columns (A column, B column, C column) are set, and as described above, from the detection result of each detection point column and the comparison result thereof, the mobile object Since T and its movement and passage are discriminated, they can be surely grasped without omission and the number of passages in a predetermined direction can be accurately counted. However, the detection point sequence is set to other numbers such as 4 or 5 along the passing direction of the moving object T, and all of them or the detection result of the detection point sequence at an arbitrary position is used. Thus, the passage of the moving body T may be determined and count processing may be performed.

  It should be noted that the detection points P of each detection point sequence are appropriately provided in necessary positions according to the size of the counting target region R, the size of the captured image G, the installation height of the camera 2, or the required counting accuracy. You only have to set it. At that time, when setting a number of 8 or less, each detection point sequence and each calculated data (see FIG. 6) can be processed by 1 byte (8 bits), and each process and calculation is performed by 8 bits. Since it can be easily performed using a CPU, it is desirable to set the number of detection points P in each detection point sequence within eight. Further, a plurality of detection points P in each detection point row are obliquely crossed to some extent in addition to the direction orthogonal to the passing direction of the moving body T, depending on the state of the counting target region R, the moving mode of the moving body T, and the like. Or may be set in another direction that intersects the passing direction, such as being arranged so as to be curved.

  Further, in the present embodiment, the number of detection points P in the detection point row is fixed, and the mobile body T is detected and counted. However, the number of detection points P set is the number of mobile bodies T at each count. It may be configured to be changeable according to the counting result or the like. In this case, for example, the number of counts per unit time is compared with a preset threshold value of the number of counts, and the passing situation of the mobile unit T is judged from the comparison result, and detection is performed according to each situation. Setting is made by changing the number of detection points P in the point sequence.

  As described above, the case where the camera 2 is installed near the door 91 of the building 90 has been described as an example. However, the counting device 1 counts, for example, the moving body T passing through the passage of the building 90 or the outdoor The camera 2 can be installed at other positions such as counting cars and people passing through the road, and can be used for counting the moving bodies T passing through various count target regions R. The present invention can also be realized as a program for causing the computer of the counting device 1 to function as each means described above executed by the counting device 1.

It is a schematic diagram which shows schematic structure of the counting device of the moving body of this embodiment. It is a functional block diagram which shows schematic structure of the count processing apparatus of this embodiment. It is a schematic diagram for demonstrating the setting of a detection point. It is a schematic diagram which shows a part of number conversion table of this embodiment. It is a schematic diagram for demonstrating the determination by the number determination part of this embodiment. It is a schematic diagram for demonstrating the data processing of this embodiment. It is a schematic diagram for demonstrating a process when a moving body is detected by the detection point row | line | column. It is a schematic diagram for demonstrating the comparison process between each detection point row | line | column and the detection point of in data. It is a flowchart which shows a part of counting process procedure by the counting device of this embodiment. It is a schematic diagram which shows the specific example of a count process with a block. It is a flowchart which shows the procedure of the count process by the counting device of this embodiment. It is a schematic diagram which shows the example of the count process when a count flag is ON. It is a flowchart which shows a part of counting process procedure by the counting device of this embodiment. It is a flowchart which shows a part of counting process procedure by the counting device of this embodiment.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Counting device, 2 ... Camera, 10 ... Count processing device, 11 ... CPU, 12 ... ROM, 13 ... RAM, 14 ... Control part, 15 ... Storage unit, 16 ... setting unit, 17 ... counting unit, 18 ... setting input unit, 19 ... input / output unit, 20 ... image processing unit, 21 ... detection unit, 22. ··· Number determination unit, 23 ··· Passing determination unit, 24 ··· Comparison unit, 25 ··· moving object discrimination unit, 90 ··· building, 91 ··· door, G ··· shot image, K ... bus, P ... detection point, R ... counting target area, T ... moving body.

Claims (9)

A moving body counting apparatus that counts a moving body that is photographed from above by a camera and counts a moving body,
Setting means for setting a plurality of detection point sequences of the moving body composed of a plurality of detection points arranged in the direction intersecting the passing direction of the moving body in the captured image of the counting target region, along the passing direction;
Detection means for detecting a moving body for each detection point of each detection point sequence from the captured image of the counting target region;
Based on the detection result of the moving body at each detection point of the plurality of detection point rows, passage determination means for determining the presence or absence of the moving body that passes through the counting target region;
Counting means for counting the number of moving objects that pass through the counting target area in the passing direction;
A moving body counting apparatus comprising: In the moving body counting device according to claim 1,
The passage determination means compares the position of the detection point that detected the moving body between the detection point rows, and the moving body that moves the counting target area in the passing direction based on the comparison result of the position. A moving body counting apparatus comprising: a determination unit; and determining presence / absence of the passing moving body based on the determination result. The mobile counting device according to claim 2, wherein
The moving body counting device characterized in that the discrimination means discriminates a moving body that moves in the passing direction based on a change in a position along the passing direction of a detection point that detects the moving body between detection point rows. . The moving body counting device according to any one of claims 1 to 3,
A moving body counting apparatus comprising: a number determining means for determining the number of moving bodies detected in a detection point sequence based on detection results at a plurality of detection points in each detection point sequence. In the counting device of the mobile object according to any one of claims 1 to 4,
The setting means is means for setting at least first, second, and third detection point sequences arranged in order along the passing direction;
The moving object count, wherein the passage determining means determines the presence or absence of the moving object passing based on the detection result of the moving object at each detection point of the first, second, and third detection point sequences. apparatus. The mobile counting device according to claim 5,
Means for discriminating a moving body that has entered the counting target region based on the detection result of the moving body for each detection point of the first detection point sequence;
Means for discriminating a moving body moving out of the counting target region based on the detection result of the moving body for each detection point of the third detection point row;
Means for comparing the detection points where the moving bodies of the second and third detection point rows are detected and determining the moving body moving in the passing direction;
Means for comparing the determination results of each means for determining the moving body,
The moving body counting apparatus, wherein the passage determining means determines the presence or absence of the moving body passing through based on the comparison result. The mobile counting device according to any one of claims 1 to 6,
The counting device of the moving body, wherein the counting means counts each moving body that passes through the counting target region in opposite directions. The mobile counting device according to any one of claims 1 to 7,
Provided with means for returning the passage determination means to the initial state at the start of the determination process and restarting the determination on the condition that a moving body is not detected continuously at all detection points in a preset number of captured images. A moving body counting apparatus.   The program for functioning a computer as each means of the counting device of the moving body described in any one of Claim 1 thru | or 8.

Images