JPH04120682A - Counter - Google Patents

Abstract

PURPOSE:To estimate the number of moving objects by calculating the number and area of the areas of the moving objects and to estimate the speeds of the moving objects by calculating the widths and heights of the areas by extracting the areas from a two-dimensional picture formed by piling up one- dimensional slit pictures in the temporal direction. CONSTITUTION:When a one-dimensional signal f(x) is inputted to a slit image input circuit 302, the circuit 302 delivers the signal to a difference circuit 304 as a two-dimensional signal f(x, t). The circuit 304 calculates the absolute value of the difference between a background picture signal b(x) and the input signal f(x, t) and draws a silhouette pictures S(x, t) by extracting changing points after binarization. The picture S(x, t) is outputted to a noise removal circuit 305 for shaping. The shaped picture S(x, t) is sent to an area labeling circuit 306 for labeling. At the same time, a statistic calculation circuit 307 calculates the 0th-, 1st-, and 2nd-order moments and feature amounts of the minimum and maximum values of the coordinates of each label area. A labeling completion detection circuit 308 outputs the feature amounts of the labels to a counter circuit 309.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to an apparatus for counting moving objects, such as people, vehicles, or objects on a conveyor belt, using image processing.

(Prior Art) Conventionally, as a means for counting moving objects such as people, vehicles, or objects on a conveyor belt, there have been methods such as manual counting, mechanical counting, phototube counting, and image processing counting. Proposed.

(Problems to be Solved by the Invention) In the conventional counting means, the manual counting means:
For example, this is carried out to measure the number of pedestrians on roads, the number of visitors to various events, traffic volume, etc., but most of these are counted manually, and there are some failures due to monotonous work, so automation is desired. ing.

In addition, counting means using mechanical or phototubes are often used, for example, to measure the number of people entering and exiting a specific venue, but this is not done by placing some kind of restriction on the walking path or by using the above-mentioned counting means. There were problems such as limited installation locations and relatively low accuracy.

On the other hand, recent advances in image processing technology have made it possible to count moving objects from moving images. The problem was that it was difficult.

(Object of the Invention) In view of the above-mentioned circumstances, an object of the present invention is to provide a counting device that can automatically measure the number, speed, etc. of moving objects with high precision and allows the installation location of a device for capturing images of moving objects to be free. purpose.

(Means for Solving the Problems) In order to solve the above problems and achieve the objectives, the present invention extracts an area corresponding to a moving object from a two-dimensional image created by stacking one-dimensional slit image signals of a certain length in the time direction. extract,
It is characterized by having an image processing unit that outputs the number of regions in an image.

The image processing unit also includes a circuit that calculates the number of regions and the area of the region to estimate the number of moving objects, and a circuit that calculates the number of regions as well as the width and height of the region, and calculates the speed of the moving object from the calculation output. It has a circuit for estimating.

(Function) The present invention does not process moving images frame by frame, but extracts a region corresponding to a moving object from a two-dimensional image obtained by stacking one-dimensional slit images of a fixed length obtained from one scanning line in the time direction. Then, the number of moving objects is estimated by calculating the area of the area along with the number of areas, and the speed of the moving object is estimated by calculating the width and height of the area along with the number of areas.

(Example) FIG. 1 shows a usage state diagram of the device of the present invention. Here, the moving object is a person 205 walking on the road 104 (104' is the edge of the road). The method can also be applied to moving objects such as objects.

In FIG. 1, 101 is a television camera TV and an A/
102 is a counting device of the present invention; 103 is a device for displaying the counting result output of the counting device; Generally, one-dimensional signals (slit images) of digital image data photographed by the video input device 101 or the like are sequentially input to the counting device 102 of the present invention.

The video input device 101 is not limited to a television camera TV.
A one-dimensional sensor may also be used. Further, the sample signal is not limited to visible light, but may also be a distance data signal using ultrasonic waves or the like. Hereinafter, visible light will be explained as an example. In the case of distance data, brightness can be read as a distance value.

To operate this, the digital image output from the video input device 101 is input to the counting device 102 of the present invention, and various feature quantities are calculated through image processing.The moving objects are counted using them, and the counting results are displayed on the display device. Output to 103.

FIG. 2 is an explanatory diagram of the operating principle of the apparatus of the present invention. In the figure, 201 is a scene, 202 is an image at a certain time that is displayed on the television camera TV in the video input device 101 shown in FIG. 1, and 203 is an object to be processed. A slit 204 for forming a two-dimensional spatio-temporal image (hereinafter referred to as spatio-temporal image) is a spatio-temporal image processed by the image processing unit of the counting device 102.

The person 205 walking in the scene 201 appears as an object image 206 on the spatio-temporal image 204. This object image 2
06 appears in response to the person 205 passing through the slit 203, and by counting the number of object images as a basic operation, the number of persons passing through can be directly determined. Furthermore, the walking direction of the person 205 (pedestrian) (a)
, (a) are different, the object image 20 on the spatiotemporal image 204
Since the time of 6 and the inclination with respect to the X axis are different, the speed can also be calculated.

FIG. 3 is a block diagram of an embodiment showing the configuration of the counting device 102 of the present invention, and in the figure, 301 is the video input device 10.
1, a signal input terminal 310 is a count result output terminal to the display device 103, and 302 to 309 constitute an image processing section. Here, 302 is a slit image input circuit, 303 is a background part setting circuit, 304 is a difference circuit between the circuits 302 and 303, 305 is a noise removal circuit, 306 is a region labeling circuit, 307 is a statistics calculation circuit, and 308 is a labeling circuit. The completion detection circuit 309 is a counting circuit.

To explain the operation of this image processing section, the video input device 10
1 to the slit image input circuit 302 via the signal input terminal 301, the slit image input circuit assigns time and space coordinates.

That is, when a one-dimensional signal f(x) is input, the clock of the device at that time is used to record the time and the two-dimensional signal f(x,
t) and is passed to the difference circuit 304. This differential circuit 30
4 calculates the absolute value of the difference between the background image signal b (x) stored in the background image memory of the background part setting circuit 303 and the input signal, and extracts the changed portions in binary form. ,
Create a silhouette image S(x, t).

That is, s(X, O= If (x, t)b(Xl, x
=0. “t-yXsIz.

Calculate. Here, Xs1□. is the size of the image.

The background image signal b (x) is inputted into the background section setting circuit 303 by a user issuing a command, assuming that no pedestrian (person 205) is present. Or, in advance,
Set when installing the device. Or the differential circuit 30
Among the signals input to Xl f(X, t)-, f(
When the value of X, t-1) is calculated and it is small, the signal at that time is set as the background in the background image memory.

Furthermore, in the processing of the labeling completion detection circuit 308,
When it is determined that there is no area being tracked, the background part setting circuit 3
A command is passed to 03 to update the background.

The silhouette image S(x,t) output from the difference circuit 304 is sent to a noise removal circuit 305, where the figure is shaped.

For shaping, the silhouette image S(x,
t) Alternatively, by providing a memory that holds the immediately previous silhouette image, the silhouette image S for two scanning lines can be stored.
(x, t-1), using S (x, t),
conduct. For example, shaping processing is often used as a method for binary images in image processing. This is achieved by including a circuit that performs enlargement/reduction processing.

The shaped silhouette image S(x, t) is sent to a region labeling circuit 306 to label the region.

At the same time, for each label area, a statistics calculation circuit 307 calculates feature quantities such as each moment of order 0.1.2 and the minimum and maximum values of area coordinates. Labels are merged when two regions are found to be connected. Features are also integrated.

The labeling completion detection circuit 308 has a memory A that stores the label of the area being scanned, and a memory B that stores the label of the scanning result of the previous time t. B, and when a certain label has been used until the previous time t-1 and is not currently used or integrated into another area, it is detected that scanning of the area of that label has been completed; The feature amount of the corresponding label at that time is sequentially outputted to the counting circuit 309.

The counting circuit 309 has a counter and an arithmetic unit, analyzes the feature amount received from the labeling completion detection circuit 308, and outputs the statistics of the moving objects and, if necessary, the speed of each moving object from the output terminal 310. Counting circuit 309 operates as follows. Among the features, the 0th moment, that is, the area is M,
When , calculate the condition as follows. That is, the area M. If it is thicker than a certain threshold T, count the number of people by dividing it by the standard area (SS), and if it is below a certain threshold T, then TI, and if it is larger than (threshold of the noise area), it is 1. People and others are marked as noise. This function 1ntO performs integer conversion. Since this number is obtained for each area, the total number can be obtained by summing up the numbers one after another. Then, it is output and displayed from the output terminal 310 to the display device 103.

Here, the counting device according to the invention described in claim (1) includes a circuit that does not use the standard area SS and only counts once for each area (pointing to the part of the figure in FIG. 2).

In addition, the standard area SS and threshold value T, which were explained above, are
The counting device according to the invention described in claim (2) includes a circuit that sequentially accumulates the sum for each area using T.

Next, an embodiment of the invention according to claim (3) will be described.

FIG. 4 is an explanatory diagram for determining the speed of a moving object, in which a cylindrical object 402 moving at a velocity V is observed with a slit 401 placed in one direction to form a spatio-temporal image. , is an example when performing counting.

However, 403 in the figure indicates the time of the object 402. indicates the position on the image. Similarly, 404 indicates the position of the object 402 on the image at time L1.

On the spatiotemporal image, the object 402 forms an image with a width and a height τ, as shown in FIG. Here, if the apparent height H on the motion image of the object 402 and the apparent direction n are known in advance, θ and τ can be found, and Equation (1)
From this, the direction of the object velocity and the angle β of the slit 401 can be calculated. Furthermore, using equation (2), the magnitude of the object velocity IV
I is required.

That is, It is.

The cylindrical object 402 is shown in the spatio-temporal image 5 as shown in FIG.
02, it looks like an object image 501. It is easy to determine the height by 12 widths from this image. Therefore, among this counting circuit and statistics, from the minimum to maximum value, τ,
After calculating h, a module is installed that uses preset angle information O and the apparent H of the object to calculate the velocity and generates a high force.

In addition to calculating speed using angle and speed as described above,
In the case of counting the number of people traveling on a road, it may be sufficient to confirm only the approximate speed and direction. If the television camera TV is properly installed in the direction of the road 104, it can be determined only by the positive or negative sign of the angle tanψ in FIG. Therefore, using the second moment calculated earlier, we find ψ by the solution of . However, calculate the inclination of the axis where the moment is larger. For example, in the installation shown in FIG. 4, the object 402 is moving to the right when tanψ is positive, and to the left when tanψ is negative.

It should be noted that, as can be easily inferred from the above embodiments, the devices of claims (2) and (3) can be constructed by combining circuits.

(Effects of the Invention) As explained above, when a one-dimensional slit moving image taken using an imaging device such as a television camera is input to the counting device of the present invention, it passes through the slit position. This method has the advantage that it is possible to count the number of moving objects that have moved, and automatic measurement is possible.

In addition, an imaging device such as a television camera can be installed freely, and the number, speed, etc. of moving objects can be automatically measured with high precision.

[Brief explanation of drawings]

FIG. 1 is a diagram showing the usage state of the device of the present invention, FIG. 2 is an explanatory diagram of the operating principle of the device of the present invention, FIG. 3 is a block diagram of an embodiment showing the configuration of the device of the present invention, and FIG. FIG. 5 is an explanatory diagram for determining the velocity of a moving object, and FIG. 5 is a diagram for explaining an object image in a spatio-temporal image. 101...Video input device, 102...Counting device,
103... Display device, 104... Road, 104'.
... Edge of road, 201... Scene, 202... Time image, 203.401... Slit, 204°502.
... spatio-temporal image, 205... person, 206°501.
...Object image, 301...Signal input terminal, 302...
Slit image input circuit, 303...Background part setting circuit, 3
04... Differential circuit, 305... Noise removal circuit, 3
06... Area labeling circuit, 307... Statistics calculation circuit, 308... Labeling completion detection circuit, 309
...Counting circuit, 310...Output terminal, 402...
Cylindrical object, 403.404...Position of the object. Comparison of the speed of the book 402 and the speed of the book 401.

Claims (3)

[Claims] (1) The feature includes an image processing unit that extracts a region corresponding to a moving object from a two-dimensional image formed by stacking one-dimensional slit image signals of a certain length in the time direction, and outputs the number of regions in the image. counting device. (2) The counting device according to claim 1, wherein the image processing section includes a circuit that calculates the number of regions and the area of the regions to estimate the number of moving objects. (3) The counting device according to claim 1, wherein the image processing section includes a circuit that calculates the number of regions as well as the width and height of the region, and estimates the speed of the moving object from the calculation output.