JP2006094293A - Difference-image detecting system and difference-image detecting method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a difference-image detecting technique detecting a moving body when no matter occurs, when an occurrence at any time of anything is difficult to conduct a detection or the like in a photographed image at all times. <P>SOLUTION: Static images 11 and 12 read in a work region by cutting out a stored dynamic image are compared by a difference-detecting software. When the difference of the images 11 and 12 is a previously specified threshold value and is less than the threshold value in the lower limit of the difference recognized as the difference between both images as the result of the comparison, a comparison processing between the image 12 and the image 13 as the next two images to be compared containing the after image 12 in a time series used for a comparison at this time is carried out. When a change can be confirmed by comparing the images 12 and 13, the comparison is transferred to the comparison of the next two images 13 and 14. When a variation larger than the threshold value in the lower limit of the previously specified difference is detected between the image 13 and the image 14, the image 14 in which the change is recognized is conserved. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a differential image detection technique for detecting a difference between images, and more particularly to a differential image detection technique for finding a moving object from continuous images recorded in time series.

  Conventionally, as a monitoring system capable of accurately tracking a target moving object even when there are a plurality of moving objects, for example, a monitoring system described in Patent Document 1 below has been known. The system described in Patent Literature 1 stores density image pattern information of each object in a time series from images captured by a monitoring camera, and compares the past information with the previous information and previous information. Is used. Thus, a moving object is detected and the moving object is tracked.

JP-A-8-205131

  In the technique described in the above-mentioned Patent Document 1, in the case where a tracked image or a constantly taken image is stored and reviewed again, for example, past data, in particular, in the same manner as a video deck fast forward / rewind operation, A method is used to search for scenes with large changes such as moving objects.

  In this case, when an image taken constantly and an image with little operation such as a landscape image continues for a long time, it is necessary to visually distinguish the image and perform a rewinding operation. Therefore, for example, when nothing happened or when it is difficult to find out what happened at what point, the task of finding a moving object has a heavy burden on the user, and it has been difficult to effectively use image data.

  An object of the present invention is to provide means for effectively reusing stored images.

  The present invention provides a means for importing a stored moving image file into an image processing means such as a computer, a means for discriminating a moving object based on a difference obtained by comparing and comparing the captured images, and an image of the discovered moving object only. Is stored as a still image file in a specific storage area. By detecting changes in the moving image file based on the specific image in the still image file saved in this way, there is no need for visual confirmation such as real-time playback of the moving image file or operation of the device after confirmation. Can be.

  The following method can be used as a difference detection method considering an image error. (1) A monitoring target frame is provided in the image, and a portion where an error is likely to occur is intentionally removed. (2) Taking an average of past images to make it difficult to detect moving object detection due to errors. (3) It has a fixed reference image, and only the true moving object is found by comparing the reference image with other target images.

  ADVANTAGE OF THE INVENTION According to this invention, the burden of the user at the time of discovering a moving body from the preserve | saved moving image can be reduced. In addition, the accuracy of discovery can be increased.

  An image difference detection system according to an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a functional block diagram showing a configuration example of an image difference detection system according to this embodiment. The image difference detection system according to this embodiment may be executed by a dedicated circuit or the like, but is generally realized by a program executed on the computer 01 that performs image processing.

  As shown in FIG. 1, a computer 01 constituting the image difference detection system according to the present embodiment is constituted by a hard disk or a removable disk, and stores an image file based on a moving image taken by a digital video camera, for example. The image file storage unit 02, a work area (a high-speed memory such as a RAM) 03 for executing a program, and a hard disk or a removable disk are selected from the moving image and selected based on the presence or absence of the difference. For example, it has a differential image file storage unit 04 that stores still images, and a CPU 6 that controls the entire image. The hard disk / removable disk 02 constituting the image file storage unit and the disk 04 constituting the difference image storage unit may be the same or stored in different storage areas. Either one of the image file storage unit 02 and the difference image file storage unit 04 or another storage unit stores an OS, application software for detecting a difference, and the like. The difference detection software 05 is expanded in the work area 03 and operates under the control of the CPU.

  FIG. 2 is a schematic diagram showing an image obtained as a result of the operation of the image difference detection system according to the present embodiment. The description will be made with reference to FIG. As shown in FIG. 2, for example, still images 11 and 12 obtained by cutting out moving images stored in the image file storage unit 02 and reading them into the work area (memory) 03 are compared by the difference detection software 05. The comparison may be made by comparing the whole image, comparing feature points in the image, luminance, chromaticity such as RGB, and saturation individually or comprehensively. For example, as will be described later, when it is known in advance that a significant change is likely to appear in a partial region on the right side of the tree in FIG. 2, a frame that defines the partial region on the right side of the tree is used as a feature point. It is also possible to provide only the image within the area as a comparison target. In addition, when it is desired to exclude a change caused by a small shake of the leaf tip of the tree (not to be stored in the difference image file storage unit 04), the change in the chromaticity (RGB) of the upper half of the image, particularly the G (green) component The change may be evaluated minutely (the threshold value is increased). As a result of this comparison processing, when the difference between the images 11 and 12 is a threshold value specified in advance and there is a difference between both images, that is, when the difference is less than the lower limit threshold value of the difference recognized as the difference (in this case, In order to determine that no change is recognized), the comparison process of the next two images 12 to be compared including the later image 12 used in the comparison in this time series and the image 13 is performed. Execute. The comparison target images are generally still images, but moving images may be compared.

  If no change can be confirmed by comparing the images 12 and 13, the execution is shifted to the comparison of the next two images 13 and 14. In this way, the comparative images are sequentially shifted. For example, when a change amount equal to or greater than a predetermined lower limit threshold value is detected between the image 13 and the image 14, the image in which the change is recognized (the latter in time series) 14 is stored in the difference image file storage unit 04.

  Further, the difference between the two images is stored in the difference image file storage unit 04 by repeatedly storing the images only when a change amount equal to or greater than the threshold value of the lower limit of the difference specified in advance is detected. Only the confirmed image remains. At this time, the image data stored in the difference image file storage unit 04 and the moving image data are associated with each other by data (referred to as association information) serving as a mark such as time data. Since the image data stored in the difference image file storage unit 04 has a very small amount of data compared to the moving image data in the image file storage unit 02, the image data stored in the difference image file storage unit 04 Image data in the image file storage unit 02 can be quickly searched using the associated time data as a search key. Even when there is no association by time data, the data is stored in time series. Therefore, the image data in the image file storage unit 02 is stored on the basis of the image data stored in the difference image file storage unit 04. The search process can be done easily. Although the difference between the two images is detected here, a process of detecting the difference for three or more images may be performed. There are two methods for detecting the difference, such as when the difference gradually changes like a natural change such as a tree shake by comparing several frames of images, or when a person suddenly appears. There is also a method of distinguishing from a case where the difference suddenly changes in a frame and treating at least one of the former difference and the latter difference as a true difference depending on the purpose. In short, this is a method for monitoring the amount of change in the difference. It is also possible to save the former difference and the latter difference in an identifiable state. By doing in this way, the difference according to the objective, such as detecting a gradual change or detecting a sudden change, can be detected and used later.

  FIG. 3 is a flowchart showing the flow of processing by the difference detection software 05. As shown in FIG. 3, the processing of the difference detection software 05 is started, one image is read from the image file storage unit 02 (step 301), and the one read image is set as a reference image (step 302). Next, another image is read from the image file storage unit 02 (step 303) and used as a comparison target image (step 304). The reference image and the comparison target image are compared (step 305). If both images match (is less than the lower limit threshold of the difference specified in advance), the comparison target image is stored in the difference image file storage unit 04. Save (step 306). Next, the image used as the comparison target image is set as a reference image (step 307). It is checked whether there is an image that has not yet been read in the image file storage unit 02 (step 308). If there is an image that has not yet been read, one image is read from the image file storage unit 02 again. (Step 303), the read image is set as a comparison target image (Step 304), and thereafter, similar comparison processing is continuously executed. If there is no image that has not yet been read in the image file storage unit 02 (if all images have been read in a certain unit), the processing is terminated.

  As described above, according to the image difference detection system according to the present embodiment, for example, a still image cut out from original image data such as a moving image is sampled in time series to compare images close in time series. The difference between the images based on the image is detected, and only the images whose difference is greater than or equal to a certain value are stored separately in association with the original image data, so that the storage location (position, Time etc.) becomes easier to search.

  In the process shown in FIG. 3, there is also a method in which the process of step 307 is omitted and the process proceeds to step 308. If the process of step 307 is not performed, there is an advantage that the process is simplified because it is not necessary to replace the reference image. Even in this case, only the true moving object can be found by comparing the reference image with other target images. However, compared with the processing that does not pass through step 307 described above, there is a possibility that the threshold value may be suddenly exceeded at a certain point in time when, for example, a slight change occurs in time series, the image immediately before that There is a possibility that it is considered a moving body suddenly even though the difference is not so large.

  In the above-described embodiment, the example in which the entire image is the comparison target as the target image as illustrated in FIG. 2 has been described. However, the frame (201, 202, 203, 204) that defines a partial region in the image. It is also possible to make only the images within the comparison object. A plurality of frames can be specified for one image. By comparing only the inside of the frame and excluding the image part other than the frame, the image part that may cause an error included in the image can be excluded from the comparison object in advance, so that more accurate comparison is possible. Is possible. For example, in the rightmost image 14 in FIG. 2, since the human image 205 suddenly appears in the frame 204, the difference from the image in the previous frame 203 is much larger in the frame than in the entire image. , Makes it easier to find differences. In addition to the concept of a frame, there is a method of extracting a feature portion of image data and detecting a difference based on comparison of the feature portion. Further, when the moving image data is data that can be predicted to some extent, it is also possible to employ a method that uses only one or a plurality of fixed reference images as reference images.

  Next, an image difference detection system according to a second embodiment of the present invention will be described with reference to the drawings. In the image difference detection system according to the first embodiment, for example, when comparing the image 11 and the image 12 in FIG. There is a possibility of judging. Such an error is taken into consideration in the image difference detection system according to the present embodiment.

  4 and 5 are flowcharts showing the flow of image difference detection processing in consideration of errors in the difference detection software 05 used in the image difference detection system according to the present embodiment. In step 302 as a reference image of the difference detection software 05 shown in FIG. 3, one image is read from the image file storage unit 02 and used as a reference image. On the other hand, in the image difference detection process according to the present embodiment, a plurality of (n−1) images are read in addition to one image (step 401), and a total of n read images are read. An average is taken (step 402), and the average image data obtained by taking the average is used as reference image data (step 403).

  Similarly, instead of step 307 (FIG. 3) for updating the reference image of the difference detection software 05, the oldest image and the comparison image are replaced with the average from the original n images averaged and used as the reference image, and the average is again performed. (Step 501), this image is set as a reference image (step 502). As described above, according to the image difference detection system according to the present embodiment, by performing a process using the average of n images as a reference image, it is possible to reduce the possibility of detecting an error caused by a leaf shake or the like. It is possible to suppress a decrease in accuracy due to an error in the moving object discovery process.

  As described above, according to the image difference detection system according to each embodiment of the present invention, it is possible to find a moving object with less burden on the user by reading an existing image into a computer and comparing them. There is an advantage.

  A more specific example of the target system is shown below. In the first specific example, captured image data in a surveillance system that always records in a convenience store or the like is stored, and there is a large movement when an incident occurs. It can be used as case investigation information by saving as an image file and referring to the image data based on the image file.

  Further, when an abnormality in an exhibit is noticed in a museum or the like, the abnormality occurrence time can be quickly investigated by searching for a moving object image related to the abnormality in the stored image.

  The above is an example, but if generalized and described, the present invention can be applied to various scenes where it is necessary to look back on an already stored image. Further, the difference image can be used as an index of a moving image file, and can be applied to a moving image file index creation technique.

  A program for causing a computer to execute each step in the present embodiment and a storage medium storing the program also fall within the scope of the present invention.

  The present invention can be applied to a monitoring system, an abnormality occurrence detection system, and the like. It can also be used for editing moving image files and creating indexes.

It is a functional block diagram which shows the example of 1 structure of the image difference detection system by embodiment of this invention. It is a figure which shows a mode that the comparison of the image was performed by embodiment of this invention. It is a flowchart figure which shows the flow of a process from the detection start of the difference based on the difference detection software used for the image difference detection system by the 1st Embodiment of this invention. It is a flowchart figure which shows the flow of a process from the detection start of the difference based on the difference detection software used for the image difference detection system by the 2nd Embodiment of this invention. It is a flowchart figure which shows the flow of the process which changed the process which makes the comparison object image in 1st Embodiment the reference | standard image.

Explanation of symbols

DESCRIPTION OF SYMBOLS 01 ... Detection apparatus (computer), 02 ... Image file storage part (hard disk or removable disk), 03 ... Storage device (memory), 04 ... Difference image file storage part (hard disk or removable disk), 05 ... Difference detection software, 11 -14 ... image, 201-204 ... frame, 205 ... human figure.

Claims (6)

An image file storage unit for storing moving image data;
A differential image detection unit that sequentially compares a reference image and a comparison target image respectively cut out from the moving image data, and detects a difference between the reference image and the comparison target image based on the comparison;
A difference image detection system comprising: a difference image file storage unit that stores the comparison target image when a difference is detected by the difference image detection unit.   Based on the comparison target image stored in the differential image file storage unit and the correlation information, by adding association information for associating the moving image data to the comparison target image stored in the differential image file storage unit The difference image detection system according to claim 1, further comprising search means for searching the image file storage unit. In a differential image detection method realized by image processing means such as a computer,
A first step of cutting out the reference image and the comparison target image based on the stored moving image data;
A difference image comprising: a second step of comparing the reference image with the comparison target image, detecting a difference between the two images, and storing the comparison target image when the difference is detected; Detection method.   In the second step, the reference image and the comparison target image are compared, and when a difference exceeding a predetermined threshold is detected in both images, the comparison target image is stored, and the two images are stored in both images. 4. The difference image detection method according to claim 3, further comprising a step of setting a comparison target image as a next reference image when a difference exceeding a predetermined threshold is not detected.   The difference image detection method according to claim 3 or 4, wherein an average value of a plurality of images acquired from moving image data is used as the reference image.   6. The difference image detection method according to claim 3, further comprising a step of detecting the presence or absence of the difference based on a change in an image for several frames.

Images