JPH09331519A - Automatic monitor equipment - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an automatic monitor equipment by which an object is discriminated with high reliability without much operation cost. SOLUTION: The automatic monitor respectively that extracts a notice image from video information obtained by an image pickup means 100 and conducts discrimination processing is provided with a plurality of extract processing means 101, 102 extracting the notice image information at a different resolution or spatial frequency from same video information, a logic processing means 103 applying logical processing extracted by each of the extract processing means, a labelling means 105 applying labelling to the notice image generated by logic processing of the logic processing means and a discrimination means 106 applying discrimination processing to the notice image generated by the logic processing. The notice images are extracted with a plurality of resolution sets from the same video image and they are logically synthesized, then the object is extracted with high accuracy and the monitor with high reliability is attained.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic monitoring apparatus for automatically monitoring a monitoring target from a picked-up image, and more particularly, it is configured so that the monitoring target can be discriminated with high accuracy by a simple device.

[0002]

2. Description of the Related Art In the field of surveillance, an area sensor such as a surveillance camera, an infrared camera, an ultrasonic sensor or the like is used to capture and discriminate a surveillance target such as a person or a vehicle, and perform processing such as tracking or counting. . For these processes, a process of extracting the monitoring target from image information of the monitoring target captured is performed.

Taking the case of a surveillance camera as an example, there is a background subtraction method as an example of a method of extracting a region of a moving object such as a person or a car as an image of interest from an image of the surveillance camera.
In this method, the target image of the target object is extracted by subtracting the background image from the camera image. With this background subtraction method, the region of the moving object to be detected can be extracted relatively easily.

However, if the object to be extracted has an area having no difference from the background image, that area cannot be detected, and the image of interest does not necessarily indicate the area of the object to be extracted.

There is also a method of integrating an image in order to generate a background to be subtracted, but in this case, there is a problem that the region of the image of interest is tailed behind the moving direction of the extraction target (Japanese Patent Application Laid-Open No. 2000-242242). 62-114064), it is not always possible to sufficiently extract the area of the moving object to be extracted.

Therefore, it is difficult to discriminate the target object from the characteristics of the detection result such as the area of the extracted image of interest, the center of gravity, the aspect ratio of the image, and the endpoints in the vertical and horizontal directions (hereinafter simply referred to as endpoints). It was

Also in the case of an infrared camera, similarly, when wearing clothes for cutting off body temperature, a wall emitting infrared rays close to the infrared rays emitted from the surface of the object to be extracted, or If an image processing method such as a background subtraction method is used when there is a road or the like, the area of the extracted target image becomes incomplete, and it becomes difficult to determine the detection target from the features of the target image.

Further, in the conventional automatic monitoring apparatus for detecting a moving object by the background subtraction method, if there is a lightning or an ON / OFF of lighting that causes a rapid image change with respect to the background image, it is erroneously detected. May occur.

Japanese Patent Laid-Open Publication No. 3-29482 discloses an automatic monitoring device for detecting a moving object while avoiding the influence of such a photographing environment.

As shown in FIG. 11, this apparatus includes a camera 31 for photographing a moving object, and a changing area extracting section for extracting a changing area of the moving object from images taken at a time interval.
32 and the boundary line with a large density difference is thinned and extracted from the captured image, and the identification number (label) is attached to each independent boundary line.
Boundary line extraction unit 33 that attaches, a polygonal extraction unit 35 that extracts a rectangular pattern or a polygonal pattern having a predetermined area or more from the thinned boundary line in the boundary line extraction unit 33,
Polygon extracted by polygon extraction unit 35 and change area extraction unit 32
Discriminating means (moving object extraction unit) 34 that integrates the changed regions extracted in step 3 to determine the changed regions included in the polygon pattern extracted by the polygon extraction unit 35 as one changed region.
And a processing mode determination unit 36 for selecting a mode in which the region of the moving object is output from the moving object extraction unit 34, a mode in which the region other than the moving object is output, or the like.

In this apparatus, the change area extraction unit 32 extracts the change area of the moving object, and the boundary line extraction unit 33 extracts the boundary line of the moving object or the background portion. The polygon extraction unit 35
A polygon having an area larger than the size of the target object is extracted from the boundary lines extracted by the boundary line extraction unit 33.
At this time, when the rectangular patterns of the boundary lines with the identification numbers overlap, the polygon extraction unit 35 extracts a polygonal boundary line region that largely includes them.

The discriminating means 34 integrates the polygon extracted by the polygon extracting section 35 and the change area extracted by the change area extracting section 32, and the change area included in the polygon pattern is defined as one moving object. Considering it as a change area, all the partial areas of the area surrounded by the polygon are integrated.

Therefore, even if a part of the moving object is dispersed and extracted as a changing area due to the influence of the photographing environment or the moving direction of the moving object, these changing areas are integrated as one moving object, The object is reliably detected.

[0014]

However, in this automatic monitoring device, the extraction processing unit 35 for rectangles and polygons is required, and the extraction is performed only to determine whether or not they are the same moving object. There is a problem that the processing scale becomes large and the cost becomes high due to the necessity of labeling processing for the boundary line.

If the boundary line extraction accuracy is poor, polygons cannot be extracted properly, and the accuracy of the discrimination processing is significantly reduced. Further, when there is an object surrounded by a boundary line having a size similar to that of a moving object, there is a problem that the target object cannot be correctly identified.

The present invention is intended to solve such conventional problems, and an object of the present invention is to provide an automatic monitoring device capable of highly accurately discriminating a target object without incurring operating costs.

[0017]

Therefore, in the automatic monitoring apparatus of the present invention, a plurality of target images are extracted from the image information of the image pickup means by a plurality of extracting means, and simple logical processing is added to the extracted images. A new target image with high accuracy is created by using, and the determination is performed using this target image.

Therefore, the accuracy of the discrimination can be improved with a simple structure, and the influence of a sudden change in the background can be reduced. Further, cost reduction can be achieved.

[0019]

BEST MODE FOR CARRYING OUT THE INVENTION The invention according to claim 1 of the present invention is an automatic monitoring apparatus for extracting a target image from video information obtained by an imaging means and performing a discrimination process, in a resolution or space different from the video information. A plurality of extraction processing means for extracting the target image in the frequency band, a logical processing means for logically processing the plurality of target images extracted by each of the extraction processing means, and a target image generated by the logical processing of the logical processing means. The labeling means for performing labeling and the discrimination means for performing the discrimination processing on the target image generated by the logical processing are provided, and the target image is extracted from the same or a plurality of video information at a plurality of resolutions. Since it is logically synthesized, the target object can be extracted with high accuracy, and highly reliable monitoring can be performed.

According to the second aspect of the present invention, each of the extraction processing means differentially processes the video information obtained by the image pickup means by the respective processing resolution or the spatial frequency, and an image obtained by this processing is obtained. , The image obtained by taking the difference between the background image of the corresponding processing resolution and the image subjected to the differential processing is binarized and output, and the processing resolution in each extraction processing means. By changing the spatial frequency or the spatial frequency, the target image can be extracted from the same video at a plurality of resolutions and spatial frequencies.

According to a third aspect of the present invention, storage means for storing the background image is provided, and the background image generated by integrating the image information obtained by the image pickup means, or the image obtained by the image pickup means is provided in this storage means. It is configured to store the background image generated by performing the process of replacing the data value of the area other than the image of interest of the obtained video information with the background image, and generate the background image using the past video information, Can be memorized.

According to a fourth aspect of the present invention, the logic processing means performs, as the logic processing, OR processing or OR processing of a plurality of target images extracted by each of the extraction processing means.
This OR process is configured to perform a concatenation process of concatenating the images of interest that are overlapped with each other, and an image obtained by combining the images extracted by the extraction processing means, or a portion that overlaps when the images are combined. An image including is extracted as a logical processing result.

According to a fifth aspect of the present invention, the labeling means performs labeling only on the image of interest newly generated or validated by the concatenation processing of the logic processing means,
The discriminating means is configured to perform the discriminating process only on the labeled target image, and it is not necessary to delete the unnecessary target image, so that the operating cost can be reduced.

According to a sixth aspect of the invention, the discriminating means discriminates the target object by using the characteristic amounts such as the area, the center of gravity, and the coordinates of the end points of the region of the image of interest which has been connected by the logic processing means. With this configuration, it is possible to perform highly reliable determination. The invention described in claim 7 is obtained by one of the extraction processing means performing an AND process on the difference between the frames of the video information obtained by the imaging means and the difference between the video information and the background image as the extracted image. The target image is output, and the other of the extraction processing means outputs, as the extracted image, the target image obtained by taking the difference between the differential image of the video information obtained by the imaging means and the differential image of the background image. With this configuration, it is possible to reduce erroneous detection due to camera shake, fluctuation of AGC of the camera device, non-detection object such as snow, or change in sunshine.

According to an eighth aspect of the present invention, in one of the extraction processing means, A which is one before the video information obtained by the image pickup means is provided.
A first background image generation unit for updating the background image by performing a process of replacing the data value of the area other than the image of interest obtained by the ND process with the background image is provided, and the differential image of the video information is provided on the other side of the extraction processing unit. And a second background image generating means for generating a differential image of the background image by integrating the above. One and the other of the extraction processing means are independent of each other.
A background image can be generated.

According to a ninth aspect of the present invention, in one of the extraction processing means, A which is one before the video information obtained by the image pickup means is provided.
A first background image generation unit that updates the background image by performing a process of replacing the data value of the region other than the image of interest obtained by the ND process with the background image is provided, and the other background of the extraction processing unit includes the first background image The background image generated by the generating means is differentiated to generate a differential image of the background image, and the background image generated by the first background image generating means is used in the other extraction processing means. Since the background image is generated, the cost can be reduced.

Embodiments of the present invention will be described below with reference to the drawings.

(First Embodiment) As shown in FIG. 1, the automatic monitoring apparatus of the first embodiment stores an image pickup means 100 (hereinafter, simply referred to as a camera) such as a monitoring camera and a background image. Storage unit 104, and a first extraction processing unit 101 for extracting a target image by subtracting a background image from the video information of the camera 100.
And subtracting the background image from the video information from the camera 100
Second extraction processing unit 102 that extracts a target image at a resolution different from that of the extraction processing unit 101, and a logical processing unit that performs logical processing such as OR processing on each target image extracted by the extraction processing units 101 and 102. 103, a labeling unit 105 for labeling a new image of interest generated by the logical processing of the logical processing unit 103, and a discrimination for discriminating the labeled image of interest using information such as its size and shape Processing unit 10
And 6.

As shown in FIG. 2, the first extraction processing unit 101 includes a BPF 201 that allows a band of a certain spatial frequency of video information obtained by the camera 100 to pass through, and background image information stored in the storage unit 104. B that passes a certain spatial frequency band
BPF from the video information input from PF202 and BPF201
A difference processing unit 203 for subtracting the video information input from 202,
A binarization processing unit 204 for binarizing the output of the difference processing unit 203 is provided.

The second extraction processing section 102 has the same configuration as that of FIG. 2, except that the BPFs 201 and 202 are the same as the first extraction processing section 102.
The only difference is that image information of a spatial frequency different from the BPF of the extraction processing unit 101 is passed.

Further, the pass bands of the BPFs 201 and 202 are specified to extract only the boundary line having a large image density difference.
So-called edge extraction can also be performed.

Data of the image of interest extracted and binarized at different spatial frequencies by the extraction processing units 101 and 102 is input to the logic processing unit 103. As shown in FIG. 3, the logic processing unit 103 includes an OR processing unit 401 that takes the OR of the image data of interest extracted by the extraction processing units 101 and 102, and each extraction processing unit 1.
An overlap determination unit 402 that identifies an image in which the target image extracted in 01 and 102 overlaps is provided.

The labeling unit 105 also includes a logic processing unit 103.
The labeling processing unit 404 that labels the image of interest identified by the overlapping determination unit 402 and the second storage unit 403 that stores the result. The discrimination processing unit 106 reads out the image of interest stored in the second storage unit 403,
Determine whether it is the target object.

As the background image stored in the storage unit 104, the data of the area other than the image of interest extracted from the video information immediately before the image information obtained by the first extraction processing unit 101 from the camera is used as the background image. , Or by integrating the video information.

By the first extraction processing unit 101 of this apparatus,
It is assumed that the images of interest extracted using a certain spatial frequency are a and b in FIG. The first extraction processing unit 101 assigns a value of 01 ₍₂₎ to the target images a and b, and assigns 00 to the target images other than the target image.
The value of ₍₂₎ is added and output to the logic processing unit 103. On the other hand, it is assumed that the images of interest extracted by the second extraction processing unit 102 using another spatial frequency are A and B in FIG. Second
The extraction processing unit 102 adds a value of 10 ₍₂₎ to the images of interest A and B, adds a value of 00 _{(2) to} the images other than the image of interest, and outputs them to the logic processing unit 103.

The OR processing unit 401 of the logic processing unit 103 performs the OR processing of these binary images and outputs the result shown in FIG. In FIG. 6, data 11 ₍₂₎ is attached to the area where the overlap occurs, and as a result of the OR processing,
As shown in FIG. 7, the overlap determination unit 402 identifies the image including the region 11 ₍₂₎ in which the overlap has occurred as one new image of interest (this is called a connection process), and the labeling processing unit 404. Attaches a label to the image of interest subjected to the connection processing. The discrimination processing unit 106 discriminates whether or not it is a target object by using the characteristic amounts such as the area, the center of gravity, and the coordinates of the end points of the region of the image of interest to which this label is attached.

As described above, in this apparatus, the labeling is performed only on the concatenated region, and only the labeled region is used as the determination target. Since the process of erasing other than the concatenated region leads to cost increase, it is not particularly performed, and those regions are ignored in labeling. When uniformly labeling each area using a general-purpose labeling processor, it is possible to ignore areas other than the areas subjected to the concatenation processing in the discrimination processing after labeling.

As described above, in the automatic monitoring apparatus of this embodiment, since the image information is processed at a plurality of resolutions and the connection processing is performed, the target object can be reliably extracted and the reliability is high. Monitoring can be done.

Further, since the image information is processed by being divided into a plurality of resolutions and spatial frequencies, it is possible to eliminate erroneous detection that occurs when the illumination changes, the camera shakes, or a moving object starts to move.

Further, the operation cost can be kept low by labeling only the image of interest that has been subjected to the connection processing and making only the image of interest that has been labeled the object of determination.

It should be noted that the logical processing unit 103 is ORed here.
Although the processing and the overlap determination processing are configured to be performed, only the simple OR processing may be performed.

The number of extraction processing units may be three or more.

(Second Embodiment) As shown in FIG. 8, the automatic monitoring apparatus according to the second embodiment has a first extraction processing unit 10 as shown in FIG.
Reference numeral 1 denotes an inter-frame difference processing unit 301 that performs inter-frame difference processing of video information obtained from the camera 100 and extracts a moving area.
And a background difference processing unit 302 that performs a difference process between a background image and an input image to extract a target image, an AND processing unit 304 that ANDs a moving region and a target image, and an AND processing unit 304.
BPF 305 that includes a background image generation storage unit 303 that generates and stores a background image based on the output of the above, and that the second extraction processing unit 102 extracts the edge image of the video information of the camera 100.
And a background edge image generation storage unit 307 that integrates the input edge images to generate a background edge image and stores the background edge image, and a background difference processing unit that subtracts the background edge image from the input edge image.
306 and.

In this apparatus, as shown in FIG. 9, the BPF 305 of the second extraction processing unit 102 extracts the input edge image,
The background edge image generation storage unit 307 integrates this input edge image to generate a background edge image. Background difference processing unit
306 subtracts the background edge image from the input edge image to extract the moving body edge image, and outputs it to the logic processing unit 103.

On the other hand, in the first extraction processing unit 101, the inter-frame difference processing unit 301 takes the inter-frame difference of the video information obtained from the camera 100. In addition, the background image generation storage unit 30
In 3, when the AND processing unit 304 obtains the image of interest by the previous AND process, the background image is generated by the process of exchanging the video information of the camera and the background image of the region other than the image of interest. At this time, the degree of replacement can be changed between 0 and 100%. The background difference processing unit 302 subtracts the background image generated by the background image generation storage unit 303 from the video information of the camera 100 and outputs the focused image to the AND processing unit 304, and the AND processing unit 304 determines the inter-frame difference and the background. The AND processing result with the difference is output to the logic processing unit 103.

The logic processing unit 103 connects the target image extracted by the first extraction processing unit 101 and the target image extracted by the second extraction processing unit 101 to generate a new target image 3.

The subsequent processing is the same as that of the first embodiment.

In the automatic monitoring apparatus of this embodiment, the fluctuation of the sunshine that causes a change on the entire screen and the AGC of the image pickup means are used.
And D by ALC (Automatic Light Control)
Erroneous detection due to fluctuations in the C component can be avoided.

(Third Embodiment) As shown in FIG. 10, the automatic monitoring apparatus of the third embodiment has a second extraction processing section.
As the background edge image generation unit 102, the background image generated by the background image generation storage unit 303 of the first extraction processing unit 101 is subjected to edge extraction processing to generate a background edge image B
Equipped with PF308. Other configurations are the same as those of the second embodiment (FIG. 8).

The operation of this apparatus is the second extraction processing unit 102.
The method is the same as that of the second embodiment except for the method of generating the background edge image in.

In this apparatus, the second extraction processing unit 102 does not have an image memory, and only the BPF having a simple structure is used, and the background image obtained by the other processing is differentiated to obtain the background edge image. Is being generated. Therefore, the configuration of the device is simplified and the operating cost can be reduced.

As the background edge image generation means in the second extraction processing unit 102, the background edge image generation storage unit 307 of the second embodiment and the BPF 308 of the third embodiment.
And the first extraction processing unit 101 does not have a background image, the background edge image generation storage unit 307 generates a background edge image, and the first extraction processing unit 101 has a background image. The BPF 308 may be used to generate the background edge image.

[0053]

As is apparent from the above description, the automatic monitoring device of the present invention is relatively inexpensive and is not affected by lightning, lighting ON / OFF, snow, rain, or the like. ,
The target object can be extracted with high accuracy, and highly reliable determination can be performed.

Further, it is possible to eliminate erroneous detection due to variations in sunshine and changes in AGC of the image pickup means.

[Brief description of drawings]

FIG. 1 is a block diagram showing an overall configuration of an automatic monitoring device according to a first embodiment of the present invention,

FIG. 2 is a block diagram showing a configuration of an extraction processing unit in the automatic monitoring device according to the first embodiment,

FIG. 3 is a block diagram showing a configuration of a logical processing unit and a labeling unit in the automatic monitoring device according to the first embodiment,

FIG. 4 is a diagram showing a focused image extraction result by the first extraction processing unit of the first embodiment;

FIG. 5 is a diagram showing a target image extraction result by a second extraction processing unit of the first embodiment;

FIG. 6 is a diagram showing an OR processing result by the logical processing unit according to the first embodiment;

FIG. 7 is a diagram showing a result of concatenation processing by the logical processing unit according to the first embodiment;

FIG. 8 is a block diagram showing a configuration of an extraction processing unit of an automatic monitoring device according to a second embodiment of the present invention,

FIG. 9 is a diagram showing a processing procedure by the automatic monitoring device according to the second embodiment;

FIG. 10 is a block diagram showing a configuration of an extraction processing unit of an automatic monitoring device according to a third embodiment of the present invention,

FIG. 11 is a block diagram showing a configuration of a conventional automatic monitoring device.

[Explanation of symbols]

 31 Camera 32 Change Area Extraction Section 33 Boundary Line Extraction Section 34 Moving Object Extraction Section 35 Polygon Extraction Section 36 Processing Mode Determining Section 100 Imaging Unit 101 First Extraction Processing Section 102 Second Extraction Processing Section 103 Logical Processing Section 104 Storage Part 105 Labeling part 106 Discrimination processing part 201, 202, 305, 308 BPF 203 Difference processing part 204 Binarization processing part 301 Interframe difference processing part 302, 306 Background difference processing part 303 Background image generation storage part 307 Background edge image generation Storage unit 304 AND processing unit 401 OR processing unit 402 Overlap processing unit 403 Second storage unit 404 Labeling processing unit

Claims (9)

[Claims] 1. An automatic monitoring device for extracting a target image from video information obtained by an image pickup means and performing a discrimination process, wherein a plurality of extraction processing means for extracting the target image at different resolutions or spatial frequency bands from the video information. A logical processing means for logically processing a plurality of target images extracted by each of the extraction processing means, a labeling means for labeling the target image generated by the logical processing of the logical processing means, and a logical processing means And a discriminating means for discriminating the image of interest. 2. Each of the extraction processing means differentially processes the video information obtained by the image pickup means in each processing resolution or spatial frequency band, and the image obtained by this processing and the corresponding processing resolution The automatic monitoring device according to claim 1, wherein the image obtained by taking a difference from the image obtained by differentiating the background image is binarized and output. 3. A background image generated by integrating the image information obtained by the image pickup means in the storage means, or the image information obtained by the image pickup means. 3. The automatic monitoring device according to claim 2, wherein the background image generated by performing the process of replacing the data value of the area other than the image of interest with the background image is stored. 4. The logical processing means, as the logical processing, OR processing of a plurality of target images extracted by each of the extraction processing means, or the OR processing, and this OR processing.
The automatic monitoring device according to claim 1, further comprising a connection process for connecting the images of interest that are overlapped by the process. 5. The labeling means performs labeling only on a target image newly generated or validly determined by the concatenation processing of the logical processing means, and the determining means determines only on the labeled target image. The automatic monitoring device according to claim 4, wherein processing is performed. 6. The discriminating means discriminates the target object by using the characteristic amounts such as the area, the center of gravity, and the coordinates of the end points of the region of the image of interest which has been connected by the logical processing means. The automatic monitoring device according to claim 4 or 5. 7. One of said extraction processing means ANDs an inter-frame difference of video information obtained by said imaging means and a difference between said video information and a background image as an extracted image.
It is obtained by outputting the image of interest obtained by processing, and the other of the extraction processing means obtains the difference between the differential image of the video information obtained by the imaging means and the differential image of the background image as the extracted image. The automatic monitoring device according to claim 1, wherein the image of interest is output. 8. One of the extraction processing means performs a processing of replacing a data value of an area other than the image of interest obtained by the AND processing immediately before the video information obtained by the imaging means with a background image. A second background image generation unit that includes a first background image generation unit that generates the background image, and the other of the extraction processing units that integrates the differential image of the video information to generate a differential image of the background image. The automatic monitoring device according to claim 7, further comprising: 9. One of the extraction processing means performs a process of replacing a data value of an area other than the image of interest obtained by the AND processing immediately before the video information obtained by the imaging means with a background image. A first background image generation unit that generates the background image is provided, and the other of the extraction processing units differentiates the background image generated by the first background image generation unit to generate a differential image of the background image. The automatic monitoring device according to claim 7, further comprising differentiating means.