JPH0869523A - Human body recognizing device - Google Patents

Abstract

PURPOSE: To accurately grasp an entry and exiting of a person into and from a room, etc., by extracting a body in motion from the difference between an image of an image signal obtained when the moving body is present and an image obtained when the moving body is absent. CONSTITUTION: An image processing part 11 consists of a differentiating circuit part 12 as an edge detecting means and a binarizing circuit part 13, and is connected to a differential binary image memory part 14. The differential binary image memory part 14 is connected to a CPU bus line 15. Then an initial memory part 16 which stores the state wherein the moving body 5 of the differential binary image memory part 14, an inter-image arithmetic part 17 which calculates the image difference between the initial memory part 16 and the differential image memory part 14 when the moving body 5 is not present, and a feature extracting circuit part 18 which extracts features from the shape of the image difference of the inter-image arithmetic part 17 are connected to a CPU bus line 15. Further, a state decision part 19 which decides the state of the extracted person 5 from the output of the feature extraction part 18 is connected to the CPU bus line 15.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a human body recognition apparatus for detecting a motion from image information of a two-dimensional image by image processing and recognizing a plurality of people based on its characteristics.

[0002]

2. Description of the Related Art A conventional human body recognition device of this type has a method of detecting the presence or absence of a human by a pyroelectric infrared temperature sensor which is a point sensor for detecting a human.

In the human body recognition device as described above, the fact that the human body temperature is higher than the ambient air temperature when a human enters the room is utilized to detect this temperature change by the pyroelectric infrared temperature sensor. Then, various controls are performed by the output of the pyroelectric infrared temperature sensor.

On the other hand, many methods of extracting a moving object from an image have been proposed, such as a background difference method of taking a difference screen between a background screen and a current screen and extracting only a moving part, or a previous screen and a current screen. There was a frame difference method by difference calculation with the screen. In these, human detection is performed on the assumption that the extracted moving body is a human.

[0005]

However, the structure of the above-mentioned conventional example has the following problems. That is,
It is possible to detect when a person comes in by using the fact that the temperature of the person is higher than the surroundings with a pyroelectric infrared sensor that is a point sensor, but when the person moves around the detection area or for a long time. There is a problem that malfunction occurs when it exists in the detection area. That is, there has been a basic problem that it is difficult to process the information of the spatially spread area by the spot sensor.

On the other hand, the background difference method and the inter-frame difference method, which are methods for extracting a moving object from an image, extract all moving objects, and include non-human objects.

In order to detect an intruder, a method of detecting even a slight movement could be sufficiently fulfilled in order to work on the safe side. Therefore, it is possible to detect the intrusion by detecting the moving object by the background subtraction method or its improved method.

However, when the object cannot be a human being, that is, when the number of humans passing through the entrance and exit is counted, or when it is desired to distinguish a moving car from a human, it is necessary to distinguish between a moving body and a human. Come on. However, in conventional moving object detection, moving objects and humans have been treated as the same thing.

Therefore, the present invention is to solve the above-mentioned problems. The presence or absence of a moving body in the detection area is extracted as two-dimensional information by the image pickup means, and a human and a moving body are identified from the extracted image by image processing. Furthermore, the purpose is to determine the human condition at that time.

[0010]

In order to achieve the above object, the present invention provides an image pickup means, an image when the moving body of the image signal obtained by the image pickup means does not exist, and an image when the moving body exists. The moving body extraction unit that extracts a moving object based on the image difference, the feature extraction circuit unit that extracts a human from the feature of the image difference extracted by the moving body extraction unit, and the human state extracted by the feature extraction circuit unit are determined. And a state determination unit that operates.

[0011]

With the above-described configuration, the human body recognition apparatus of the present invention has two-dimensional images (visible image in the case of visible light, infrared image in the infrared region, thermal image in the case of temperature distribution, etc.) in the detection region. Image), the moving object is extracted by calculating the difference between the current image and the background image, and the feature extraction circuit unit extracts a human from the size and moving speed of the extracted moving object,
The state determination unit determines the human state.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the accompanying drawings. FIG. 1 is a configuration diagram attached to an air conditioner in one embodiment of the present invention. In the figure, 1 is a room, 2 is a wall surface, air-conditioning means 3, 4 is furniture placed in the room 1, 5 is a person existing in the room 1, 6 is a room 1 and a person 5.
Image pickup means for extracting (moving body 5 before being judged as human), 7 is an optical lens, 8 is an image pickup device, and a video signal processing unit 9
Is connected to the image processing unit 11 via the A / D conversion unit 10.
Connected to The image processing unit 11 is composed of a differentiating circuit unit 12 and a binarizing circuit unit 13 as edge detecting means, and is connected to a differential binary image memory unit 14. The differential binary image memory unit 14 is connected to the CPU bus line 15. A differential binary image memory unit 1 is provided in the CPU bus line 15.
4 includes an initial value memory unit 16 that stores a state in which the moving object 5 of No. 4 does not exist, and an inter-image calculation unit 17 that calculates an image difference between the initial value memory unit 16 and the differential binary image memory unit 14 when the moving object 5 exists. A feature extraction circuit unit 18 for extracting features from the image difference form of the inter-image calculation unit 17 is connected. Also, C
In the PU bus line 15, a state determination unit 19 that determines the state of the human 5 extracted by the output of the feature extraction circuit unit 18,
It is connected to a control unit 20 that controls the air conditioning unit 3 by the output of the state determination unit 19. And CPU bus line 1
Reference numeral 5 is configured to be controlled by the CPU circuit section 21 which controls arithmetic processing. The image processing unit 11, the differential binary image memory unit 14, the initial image memory unit 16, and the inter-image calculation unit 17 constitute a moving body extraction unit 22 for extracting moving objects.

The operation of the above configuration will be described with reference to the flowchart of FIG. In step 101, the detection of the indoor 1 situation is started, and in step 102, the image of the indoor 1 is captured by the image capturing means 6. That is, the situation of the room 1 is captured by the image sensor 8 as a two-dimensional image. The output of the image pickup device 8 is subjected to predetermined video signal processing by the video signal processing unit 9 in step 103, and then the analog video signal is converted into a digital value by the A / D conversion unit 10 in step 104, and the digital converted video signal. Is input to the image processing unit 11 in step 105.

In the image processing unit 11, in order to make the shape of the state of the room 1 clear as preprocessing, and in order not to generate unnecessary noise, the differentiating circuit unit 12 extracts a portion having a sharp change in shading. The state of 1 is extracted. Then, in step 106, the image of the room 1 image-processed by the differentiating circuit unit 12 is binarized at a predetermined level by the binarizing circuit unit 13, and the binarized image data is stored in the differential binary image memory unit 14. Remember. The preprocessed image is stored as described above.

The differential binary image memory unit 14 is connected to the CPU bus line 15, is controlled by the CPU circuit unit 21, and is connected via the CPU bus line 15 in step 107. It is determined whether or not the data is the initial value data in the state where the human 5 does not exist. For example, the data stored continuously several times is subjected to difference calculation by the inter-image calculation unit 17, and if there is no change in the difference image, it is regarded as initial value data in step 108 and the initial value memory unit 1
Store in 6.

On the other hand, if it is not initial value data, step 1
In step 09, the image difference between the image of the initial value memory unit 14 stored in step 108 and the image of the differential binary image memory unit 14 being measured at a predetermined time interval thereafter is stored in the inter-image calculation unit 17. Then, in step 110, the inter-image calculation unit 1
The motion of the room 1 (there is a motion if there is an image difference) is extracted from the images calculated in 7 by the moving body extraction unit 22, and the feature extraction circuit unit 18 extracts the human 5 from the image difference. A feature (here, a size equal to or larger than a certain size equivalent to a human being) is extracted. If it is difficult to recognize with this feature, the feature extraction circuit unit 18 determines whether or not the shape is equivalent to that of a human as compared with the shape of a human assumed from the installation position of the imaging unit 6. With the above, a human is extracted.

Next, at step 111, the feature extraction circuit unit 1
Based on the output of 8, the state determination unit 19 determines the states such as position and number.
Determine with. Next, in step 112, the output of the state determination unit 19 controls the capacity and the wind direction of the air conditioning unit 3 via the control unit 20 based on the number of people, the position, the way of movement, and the like.

Here, the method of obtaining the differential of the image in advance for extracting the moving object and then extracting the moving object by the inter-image calculation has been described. However, the moving object is extracted by performing the inter-image calculation without preprocessing the images. It doesn't matter.

Now, the image processing for detecting the state will be described with reference to FIG. FIG. 3A is a schematic diagram of the differential binary image in the state where the human 5 does not exist as the initial value of the room 1. FIG. 3B is a schematic diagram of a differential binary image when a person 5 is present in the same room 1, and FIG.
The image difference between B and B is calculated by the image calculation unit 17 and the human 5 is extracted. Here, by calculating the area of the differential binary image of a person who is the detection target of the differential binary image in C, it can be seen that a person exists in the room 1 if the area has a certain level or more. In addition, it can be understood that there are a plurality of humans when the area calculated above is twice or three times as large as the area of one person. In addition, since the room is detected at regular intervals, when there is no person left for a moment and the person returns immediately, it is possible to perform air conditioning without turning off the air conditioning means according to an instruction from the CPU circuit unit 21. It is needless to say that the differentiating circuit section that is executed in the image processing section as the edge detecting means includes a first-order differential operator, a second-order differential operator, and various other means, and any means can be used as long as it can produce the same effect. Yes.

With the above operation, it is possible to easily detect the presence of a person because it is possible to recognize the state of the room with a two-dimensional image and to extract only the person by the image difference between the case where the person exists inside the room and the case where the person does not exist inside the room. be able to.

Then, various devices can be controlled by recognizing the number, position, movement state, etc. of humans present in the room.

Further, when there are a moving vehicle and a person running from the inside, it is possible to recognize only the person from the size and the movement and count the number of the persons.

[0023]

As described above, according to the human body recognition apparatus of the present invention, the following effects can be obtained.

(1) Instead of detecting the detection area as one-point information such as detection by an infrared temperature sensor as a point sensor, the moving body is extracted by visualization or two-dimensionalization, and the characteristics (shape or shape) of the moving body are extracted. Since people are recognized based on their size, etc., it is possible to accurately grasp the entrance and exit of a room.

(2) Also, the number of persons can be detected by judging the extracted human state, and for example, the capacity of the air conditioning means can be controlled according to the number of persons, or the number of visitors can be counted.

(3) Furthermore, when it is desired to distinguish a moving car from a human, it is possible to distinguish a human from a moving body other than a human.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of a human body recognition apparatus according to an embodiment of the present invention.

[Fig. 2] Flow chart of the device

FIG. 3 is an image schematic diagram showing an indoor state for detecting the same state.

[Explanation of symbols]

 5 human 6 image pickup means 18 feature extraction circuit section 19 state determination section 22 moving body extraction section

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location G06T 7/20 H04N 7/18 D

Claims (1)

[Claims] 1. An image pickup means, and a moving body extraction section for extracting a moving object by an image difference between an image when the moving body does not exist and an image when the moving body exists in the image signal obtained by the image pickup means. A human body recognition apparatus having a feature extraction circuit unit that extracts a human from the features of the moving body extracted by the moving body extraction unit, and a state determination unit that determines the state of the human extracted by the feature extraction circuit unit.