JPH0823882B2 - Passerby counting device and sales processing device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION (a) Field of Industrial Application The present invention relates to a passerby counting device that counts the number of passersby by an image taken by a camera, and a sales processing device that processes sales based on this data. Regarding

(B) Conventional Technology As a passerby counting device for counting the number of passersby, conventionally, a device using ultrasonic waves and a device using an infrared sensor have been put into practical use.

(C) Problems to be Solved by the Invention However, in the case of using ultrasonic waves, it is impossible to identify when humans pass by in an overlapping manner. It was necessary to narrow the passage so that it would pass. Further, the infrared ray sensor is used to detect infrared rays emitted by humans, but since the infrared rays are weak, the sensitivity of the sensor must be increased, and maintenance is not easy. Further, in order to detect a plurality of pedestrians who pass by at the same time, it is necessary to provide a sensor array that crosses the passage, which is a high cost.

An object of the present invention is to provide a passer-by counting device that solves the above-mentioned drawbacks by using a camera.
An object of the present invention is to provide a sales processing device that statistically processes the number of passersby and sales by using this passerby counting device.

(D) Means for Solving the Problem The pedestrian counting device of the present invention extracts a camera installed from above the passage to just below and a moving object moving in the passage from an image captured by the camera. Moving object extracting means, moving object determining means for determining whether or not the moving object is a human when the moving object passes through a strip-shaped area that crosses a passage set in an image captured by the camera, A moving direction extracting unit that extracts a moving direction of the moving object when it is determined that the moving object is a human; and a unit that counts the extracted human image in a passage in the image data for each moving direction. The moving object determination means extracts only a circular moving object from the moving objects extracted by the moving object extraction means,
It is characterized in that it is means for determining whether or not the moving object is a human by performing processing such as pattern matching only on a moving object having substantially the same size as a human head.

Further, the sales processing device of the present invention includes the passer-by-counter device, information including the number of passers-by counted by the pass-by-counter device, past or present sales data, weather data such as weather, temperature, and humidity. And a statistical predicting means for performing statistics or prediction including relations between passers-by and sales, passers-by and weather, sales and weather, and the like.

(E) Operation In the passer-by-counter device of the present invention, the moving object is extracted from the image taken by the camera installed from above the passage to right below. When the moving object passes through a belt-shaped area that crosses the passage set in the image captured by the camera, the moving object determination means patterns only for a circular moving object having substantially the same size as a human head. By performing processing such as matching, it is determined whether or not the extracted moving object is a human. When it is determined that the moving object is a human, the moving direction extracting means extracts the moving direction and counts the moving directions. Therefore, the number of people passing through the passage can be obtained for each moving direction, and from this result, for example, the situation of entering and leaving the store, the movement of the person in the passage, and the like can be easily grasped. In addition, since the human head is extracted and counted from the upper part of the passage to the position just below, even if the passage is congested, the heads do not overlap each other, so that the human can be accurately counted. Further, even if the hair length is different for each individual, it is circular when the human head is photographed from directly above. Therefore, when the moving object passes through a strip-shaped area that crosses the passage, a circular moving object is extracted, and only from the extracted circular moving object to an object that is approximately the same size as the human head. Since it is configured to determine whether or not a person is a person by performing pattern matching or the like, this determination is not performed at all for a moving object that is obviously not a human such as a square. Further, even if the moving object is circular, the same moving object is not judged a plurality of times, so that the number of judgments can be reduced and the processing time can be shortened.

Further, in the sales processing device of the present invention, the number of passers-by counted by the pass-by counting device, past or present sales data, and information including weather data such as weather, temperature, humidity, etc. are fetched and passed by. Performs statistics or forecasts including relationships between sales, passers-by and weather, sales and weather, etc.

(F) Example An example of the present invention will be described with reference to the drawings. This embodiment is a sales statistics device equipped with an entrance / exit counter. The entrance / exit counting device detects an entrance / exit from an image of a camera installed at each entrance (passage), and counts this. Further, the sales statistics device is a device that inputs weather data and the like in addition to the number of people entering and leaving in real time to perform sales statistics and forecasts.

FIG. 1 is a block diagram of the sales statistics device of the present invention, FIG. 2 is a diagram showing a camera installation form of the sales statistics device, and FIG. 3 is a structure of a main body of an entrance / exit counting device of the sales statistics device. FIG.

The entrance / exit counting device consists of a camera 1, a main body 2 and a monitor 3.
It consists of The main body 2 is connected to the computer 4. The computer 4 also has a monitor 5 and a POS
The terminal device 6 and the modem 7 are connected. Computer 4
Is based on real-time number of visitors, number of exits, real-time sales data from POS6 and online data from modem 7 Make sales statistics and forecast future sales. As the online data, weather data, weather forecast data, stock data, etc. provided via a telephone line are used.

The camera 1 is provided with four cameras 1a to 1d. These four units are installed at the entrances and exits of different places from directly above (see FIG. 2), and passersby at each entrance and exit are photographed. The video output of this camera is input to the main body 2.

A detailed block diagram of the main body 2 is shown in FIG. Camera 1
Is connected to the input unit 20. The input units 20 are provided for the number of cameras. The input unit 20 is a buffer that buffers image signals. The image signal buffered in the input unit 20 is input to the image processing unit 21. Image processing unit 21
Are also provided for the number of cameras. The image processing unit 21 A / D converts the analog image signal. One screen of the A / D converted image data is divided into 512 × 512 pixels, and the gradation of each pixel is represented by 8 bits (256 steps). Further, the image processing unit 21 filters the noise component (quantization error or the like) included in this image data. The filtering process includes averaging process and the like. The averaging process is a process of reconstructing each pixel as an average value of pixels adjacent to it. That is, the density of the pixel P (x, y) before processing is L (x,
y), and the density after treatment is D (x, y), D (x, y) = (L (x-1, y-1) + L (x-1, y) + L (x-1 , y + 1) + L (x, y-1) + L (x, y) + L (x, y + 1) + L (x + 1, y-1) + L (x + 1, y) It is represented by + L (x + 1, y + 1) /9.This image data is input to the control unit 22 in a time-divisional order in the order of the cameras 1a to 1d. The control unit 22 is composed of a microcomputer. Then, the image data of each camera is sequentially captured, the moving body is detected from the image data, and it is determined whether or not the moving body is a human.
Whether to enter or leave the store is determined based on the moving direction, and the number of storage areas is added. The counting result is processed by the multi-image control unit 23, displayed on the monitor 3, and output to the computer 4 via the interface 24.

FIG. 2 is a diagram showing an installation form of the camera 1. The camera 1 is installed directly above the doorway 10 and faces downward so that the entire width direction of the doorway 10 can be photographed. Those who pass here are photographed from directly above. The image taken is the fifth
It becomes as shown in FIG. 6 and FIG. When displaying the image of one camera on the monitor, the display is as shown in Fig. 5,
When displaying images from multiple cameras on a single screen
It becomes like the figure.

FIG. 4 is a flow chart showing the operation of the control unit 22. This operation is performed for each camera image at regular intervals. An area for counting the number of visitors and the number of exits is provided for each camera. Further, there is one area for counting the number of customers staying in the store.

First, whether or not the object is a moving object is detected at n1. If there is no moving object, proceed from n2 to n11. If there is a moving object, it is judged whether or not it is a human (n3). If not a human, proceed from n4 to n11. In the case of a human, the moving direction is detected (n5), and in the case of entering a store, the store register corresponding to the camera is incremented by 1 (n7), and the stayer register is incremented by 1 (n8). In the case of exit, 1 is added to the exit register corresponding to the camera (n9), and 1 is subtracted from the stayer register (n10). After this, the display update operation (n
Execute 11) and return. The display update operation of n11 is
This is an operation of displaying the image data input from the image processing unit 21 on the monitor 3 and displaying the time and the number of people entering and leaving at that time as a telop.

As a method for detecting a moving object in n1, for example, there are the following methods.

(1) A method of comparing the current image data captured from the camera with the immediately preceding image data stored in the memory and detecting a moving object based on the deviation of the data. (2) What is the moving object? The image data in a state in which there is no state is stored in advance, this image data is compared with the image data captured from the camera to detect a different portion, and the moving direction is detected in the method n5 of detecting a moving object. It suffices to find the barycentric coordinates of the moving object and obtain the moving direction based on the change in the barycentric coordinates.

Further, as a method of discriminating a human from an image of a camera installed directly below the ceiling, the human characteristic may be extracted. In this case, the head is extracted as the first characteristic point. Since the human head is an ellipse, first, a circle is separated and extracted from a moving object. Here, the area of the circular object is obtained, and those obviously smaller or larger than the size of the head are excluded. Next, processing such as pattern matching is performed within a certain range from the center of the circular portion to distinguish between humans and other objects (for example, balls, balloons, etc.) and determine only humans. Count the number of humans that have been identified.

Images are taken with a camera from above the passage to just below, and the human heads are extracted and counted. Therefore, even if the passages are congested, the heads do not overlap with each other, so that the humans can be accurately counted.

Further, even if the hair length is different for each individual, it is circular when the human head is photographed from directly above. Here, a circular object is extracted from the extracted moving object, and only the object having the same size as the human head is subjected to pattern matching etc. from the extracted circular moving object to be a human. Since it is configured to determine whether or not it is, this determination is not performed on a moving object that is clearly not a human such as a square. Therefore, the number of determinations as to whether the moving object is a human is reduced, and the processing time can be shortened.

It should be noted that the detection of the moving object and the determination of the moving direction need to be performed by using the data of the entire image, but the determination as to whether the moving object is a human may be performed only in a specific range. That is, a band-shaped area that crosses the passage may be set in the image data, and when the moving object passes through the band-shaped area, it may be determined whether or not the person is a human. By doing so, it is possible to reduce the number of determinations as to whether or not the person is a human, and it is possible to improve the processing speed.

By using the above entrance / exit counting device, it is possible to count the number of entrance / exit at each entrance in real time. Further, the number of passers-by can be counted not only in the entrance and exit but also in a passage.

By statistically processing the number of stayers and the number of people entering and leaving the entrances and exits thus counted, it is possible to effectively set the in-store layout and the like.

The number of people entering and leaving and the number of stayers detected by the above operation are input to the computer 4. In computer 4,
In addition to inputting real-time stayer information, various kinds of information such as sales information for each product type from the POS terminal device 6, weather information for each time from the modem 7 and future weather forecast are accepted. It is also possible to input other event information from the keyboard of the host. From such various information, the host computer 4 processes the number of people entering and leaving and sales statistics, management processing of the number of people entering and leaving by time and the number of people staying in the store, management processing of the number of people entering and leaving according to event content, weather and events. Performs various processing such as sales forecast.

Further, since the entrance / exit can be automatically managed in real time, the above processing can be performed in real time.

The passer-by counting device of the present invention can be used not only to count the number of people who enter and exit, but also to monitor the entry of people into the exclusion zone, and
It can also be used to count the number of people waiting for an elevator or the like.

(G) Effect of the Invention As described above, according to the passer-by counting apparatus of the present invention, the camera is photographed from above the passage to right below, and the human head is extracted and counted, so that the passage is congested. However, since the heads do not overlap, humans can be accurately counted. In addition, if a circular object is extracted from the extracted moving object, and if the extracted circular moving object is a person having a size substantially the same as that of a human head, pattern matching or the like is performed to determine whether the person is a human. Since it is configured to determine whether the moving object is a square or the like, the determination is not performed on an apparently non-human moving object. Therefore, the number of determinations as to whether the moving object is a human is reduced, and the processing time can be shortened. Further, since it is determined whether or not the moving object is a human when the moving object passes through the strip-shaped area that crosses the passage, the determination is not repeated for the same moving object. The processing time can be shortened. Further, since the entire passage can be covered only by installing the camera, there is an advantage that the installation is easy and the maintenance is easy. Also, because the counting is performed according to the moving direction of the human,
For example, it is possible to easily grasp the status of entering and leaving the store.

Further, according to the sales processing device of the present invention, the number of pedestrians and past or present sales and information including weather data such as weather, temperature, humidity, etc. are taken in to obtain pedestrians and sales, pedestrians and weather, sales. It is possible to make a system and forecast of relations such as weather conditions.

[Brief description of drawings]

FIG. 1 is a block diagram of the sales statistics device of the present invention, FIG. 2 is a diagram showing a camera installation form of the sales statistics device, and FIG. 3 is a structure of a main body of an entrance / exit counting device of the sales statistics device. FIG. 4, FIG. 4 is a flowchart showing the operation of the control unit 22, and FIGS. 5 and 6 are views showing images on the monitor of the entrance / exit counting apparatus. 1 ... Camera, 2 ... Main body, 3,5 ... Monitor, 4 ... Computer, 6 ... POS terminal device. 10 …… Gateway (passage), 11 …… Passers, 21 …… Image processing section.

Claims (2)

[Claims] 1. A camera installed from above an aisle to directly below, a moving object extracting means for extracting a moving object moving in the aisle from an image captured by the camera, and an image captured by the camera. Moving object determining means for determining whether or not the moving object is a human when the moving object passes through a strip-shaped area that crosses the defined passage; A moving direction extracting unit for extracting a moving direction; and a unit for counting the extracted human images in the passage in the image data for each moving direction, wherein the moving object determining unit is the moving object extracting unit. Of the extracted moving objects, only circular moving objects are extracted,
A passer-by counting device, which is means for determining whether or not the moving object is a human by performing processing such as pattern matching only on a moving object having substantially the same size as a human head. 2. The passer-by-counter according to claim 1, and information including the number of passers-by counted by the pass-by-counter, past or present sales data, weather data such as weather, temperature, humidity and the like. A sales processing device comprising: a statistical prediction unit that captures and makes statistics or predictions including relationships between passers-by and sales, passers-by weather, sales-weather, and the like.