AU2001278180A1

AU2001278180A1 - A method of searching recorded digital video for areas of activity

Info

Publication number: AU2001278180A1
Application number: AU2001278180A
Authority: AU
Inventors: James Carrol Myers; Jeffrey J. Orth
Original assignee: COMTRAK TECHNOLOGIES LLC
Current assignee: COMTRAK TECHNOLOGIES LLC
Priority date: 2000-10-13
Filing date: 2001-06-28
Publication date: 2002-04-22
Also published as: MXPA03003268A; WO2002032129A1; CA2425855A1; EP1325630A1; CA2425855C; BR0114570A; US6434320B1

Description

A METHOD OF SEARCHING RECORDED DIGITAL VIDEO FOR AREAS OF ACTIVITY

Technical Field This invention relates to video image processing, and more particularly, to a method of searching recorded video images to quickly locate areas of activity recorded on the video. Background Art

In video recording systems, for example those use for surveillance of a premises, a camera obtains video images of a scene many times a second. These images are processed in by an image processing system and the results stored bt a digital video recorder (DVR) or the like. The images are stored sequentially; i.e., the first image obtained first, the second image obtained second, and so forth. The images are usually digitally converted analog signals, and these digital images are stored in a computer memory or on a disc. Most surveillance systems are used at times when little activity is expected within the scene viewed by the camera. Accordingly, the recorded video will typically have lengthy segments where there is no change in the recorded image. A drawback in conventional recording systems, is that in order to view those segments in which activity has occurred and been recorded, it is necessary to move through the images in their sequence of recording. That is, the sequence of recorded images stored in a memory is queued into a computer display input. The operator then sets a threshold value and pushes a PLAY button. The computer then serially sequences through the stored images until it reaches one where the amount of activity captured on the image exceeds the threshold setting. It then commences play of images beginning at that location. Even with a rapid fast viewing technique such as this, it can take an inordinate amount of time to get to a segment where there is activity which may be of interest to the viewer. The present invention solves this problem by a method that allows recorded video to be rapidly scanned with the scanning stopping only at those images where a change in the activity recorded exceeds a predetermined threshold. Summary of the Invention

Among the several objects of the present invention may be noted the provision of a method of scanning recorded video images of a scene to rapidly locate those images in which there is a predetermined amount of activity. The method involves first obtaining images of a scene over a period of time. One of the images is designated a reference image of the scene and the other images are compared to this reference one at a time. A difference image is created for each comparison of an image to the reference image, and the difference images are stored in a memory of the system. A threshold test is applied to each difference image as a measure of the activity occurring in the scene at the time the image is obtained and the results of the test are stored with the image. When the images are to be subsequently viewed, a viewing threshold related to the amount of activity occurring in the scene is established by the viewer. The test results for each stored image are then scanned to determine which images meet the viewing criteria. Those images that do are either displayed, in time sequence, for viewing or used to create a separate file for analysis. Since the viewing threshold criteria can be rapidly determined for a large number of stored images, it is not necessary to view the images in their sequential order which saves a substantial amount of time and allows the user to quickly locate pertinent information related to activities at a monitored site. Other objects and features will be in part apparent and in part pointed out hereinafter. Brief Description of Drawings

In the drawings, Fig. 1 is a simplified block diagram of an image processing system;

Fig. 2 illustrates the comparison of a reference video image to another video image to produce a difference image; Fig. 3 illustrates a threshold detection and counting portion of the system;

Fig. 4 is a simplified representation of the processed image contents stored by the system; and, Fig. 5 illustrates an image file assembled from all the stored images in which a threshold count value is exceeded.

Corresponding reference characters indicate corresponding parts throughout the drawings. Best Mode for Carrying Out the Invention Referring to the drawings, Referring to the drawings, an image processing system of the present invention is indicated generally 10. A camera 12 obtains video images of a scene and provides as an output an analog signal A. This analog signal is supplied as an input to an analog- to-digital converter 14 which converts signal A to a digital signal D. The digital signal is supplied to a signal processor 16 which processes the signal in a manner described hereinafter. After processing, the video images are stored in a memory 18 which can be a computer memory, a CD, or other data storage media. Those skilled in the art will understand that while system 10 is shown to include a video camera 12, other imaging devices, for example, an infrared (IR) detector can also be used to obtain images which are processed in accordance with the method of the present invention.

As shown in Fig. 1 , image processing system 10, over time, captures and stores a series of images l*ι-l_N. One of these images is periodically designated a reference image RI by the system. Each reference image represents a view of the scene at a given point in time. Images SI obtained subsequent in time to the reference image are compared with the reference image by processor 16. This comparison is done to determine if there has been any significant change to the scene being observed. In a security system, for example, on result of the comparison is to ascertain if an intrusion of the premises being monitored has occurred.

As shown in Fig. 2, processor 16 includes a comparator 20 having as one input the reference image RI and as a second input, a subsequent image SI. The two images are of the same size; that is, they have the same number of pixels arranged in an identical format of M rows and N columns. Because of this, comparator 20, can readily compare the two images to determine if there is any difference between them. The comparator does this by first comparing the pixel values represented by the data bits at locations RI, pι,ι and SI, p-ι_,-|. Next the comparator compares the pixel values represented by the data bits at RI, p _ι2 and SI, p_{1 ι2}, then RI, p_{1 3} and SI, pι,₃, etc., through the pixel values represented by the data bits at RI, p ,N and SI, P ,N* Comparison is performed by subtracting the one pixel value from the other, with the result being stored in a matrix representing a difference image Dl. If there is no change in the pixel values for a given location between reference image RI and subsequent image SI, a 0 value is stored in the appropriate address of the difference image matrix. Otherwise a value representing the difference is entered at the matrix location for the pixel values which have been compared.

Referring to Fig. 3, once a difference image Dl is created by comparator 20, the difference image is supplied as an input to a threshold detector and counter 22 of processor 18. Each pixel value comprising the difference is now measured against a predetermined threshold value. The number of pixel values which equal or exceed the threshold are counted, and the count value is entered a header H of what is now a processed difference image Dl'. The processed difference image is stored in image storage 18 together with the other processed difference images. Fig. 4 is a simplified representation of a stored image file F1. The contents of file F1 are each processed difference image DI DI'N stored in memory 18. Besides the contents of the image, each difference image also includes a header H which indicates the position of the image in the image processing sequence and the location of the subsequent image SI in memory 18 to which the difference image corresponds, and the count value for the number of image segments which exceeded the threshold. Now, when the stored images are viewed on a monitor 24, rather than having to scan each recorded image, in sequence, to detect activity, the contents of the difference image headers are scanned to determine which images have threshold count values exceeding a predetermined minimum. Those skilled in the art will understand that even if there is no discernible difference between a reference image and a subsequent image compared to it, there will be some differences due to system noise, etc., and some of these differences may be significant enough to exceed the threshold and thus be counted. However, these will be few in number compared with those in which there is activity within the frame.

Using the method of the present invention, once a threshold level control 26 on monitor 24 is set by the user, a FAST FORWARD control 28 facilitates rapid scanning of the images stored in memory 18. If image DI'_M is the first image in which the count threshold is exceeded, the system will rapidly access that image which is then displayed on screen 30 of the monitor. The operator then pushes a PLAY control 32 and all of the sequential images in which the threshold is exceeded are then displayed. In Fig. 4, this is represented by images DI'_M to DI'_M+x* It will be understood by those skilled in the art that the value set by the user can be changed from one time to another and the value set by the user at any one time determines the number of images which will be displayed. That is, if the viewing threshold is set to a lower value, then a larger number of stored images will meet the viewing criteria than if the viewing threshold is set to a higher value. Alternatively, and as shown in Fig. 5, the stored image file F1 can be processed to produce a file F2, the contents of which include only those stored images whose threshold count exceeds a predetermined value. Thus, prior to viewing the file, when a user sets a threshold count value, processor 18 scans all of the stored files, locates those images which meet the count criteria, and assembles the file F2 with all of the selected images appearing in their time sequence. This file can now be viewed on monitor 24, or a separate memory file containing only those images is created for separate processing.

What has been described is a method of scanning recorded video images of a scene to locate those images obtained over some period of time and in which there is a predetermined amount of recorded activity. One image is a reference image of the scene and other images are compared this reference. Difference images are created for each comparison and these are stored in a memory. When the images are to be viewed, a threshold test is applied to each determine if the amount of activity in an image is significant or not. Only those images meeting the test are then displayed, or a separate file including only those images is created.

In view of the foregoing, it will be seen that the several objects of the invention are achieved and other advantageous results are obtained. As various changes could be made in the above constructions without departing from the scope of the invention, it is intended that all matter contained in the above description or shown in the accompanying drawings shall be interpreted as illustrative and not in a limiting sense.

Claims

Claims;

1. A method of rapidly reviewing recorded video images of a scene to locate those images in which a predetermined amount of activity has occurred, comprising: obtaining a plurality images of a scene over a period of time, one of the images being a reference image of the scene; comparing the reference image with other images of the scene and creating a difference image for each comparison; performing a threshold test on the contents of the difference image to determine the number of difference image contents exceeding a predetermined threshold and counting that number; storing the difference image and threshold test results in a memory; and, subsequently viewing selected stored images including: setting a value representative of the amount of activity occurring in an image before it is displayed for viewing; reviewing the stored images and identifying those having threshold test results which equal or exceed the value; and, displaying the identified images, setting of the value and reviewing the threshold test results for the images allowing a large number of images to be rapidly reviewed without having to view the images in their sequential order thereby saving a substantial amount of time while allowing the user to quickly locate pertinent information related to activities at the scene.

2. The method of claim 1 wherein the identified images are displayed in their temporal sequence.

3. The method of claim 1 further including creating a separate file of the stored images identified as having threshold test results which equal or exceed the value.

4. The method of claim 3 wherein the images stored in the separate file are also stored in their temporal sequence.

5. The method of claim 1 wherein images of the scene are obtained using an imaging means including a camera operating in the visible portion of the light spectrum.

6. The method of claim 4 wherein the imaging means includes an infrared camera.

7. The method of claim 5 wherein the images obtained from the viewing means are analog signals and the method includes converting the analog signals to digital signals.

8. A image processing system for monitoring a scene comprising: imaging means for obtaining a plurality images of the scene, one of the images being a reference image of the scene and each image obtained being represented by an analog signal; converter means converting the analog signals to digital signals; signal processor means including a comparator comparing the reference image with other images of the scene and creating a difference image for each comparison, means performing a threshold test on the contents of the difference image to determine the number of difference image contents exceeding a predetermined threshold and counting that number, and memory means for storing the difference image and threshold test results; and, viewing means by which selected stored images can be subsequently viewed, the viewing means including means for setting a value representative of the amount of activity occurring in an image, the stored images being reviewed to identify those whose threshold test results equal or exceed the value, and a monitor for displaying the identified images, setting of the value and reviewing the threshold test results for the images allowing a large number of images to be rapidly reviewed without having to view the images in their sequential order thereby saving a substantial amount of time while allowing the user to quickly locate pertinent information related to activities at the scene.

9. The system of claim 8 wherein viewing means displays the identified images in their temporal order.

10. The system of claim 8 further including creating a separate file of the stored images identified as having threshold test results which equal or exceed the value.

11. The system of claim 10 wherein the images stored in the separate file are also stored in their temporal sequence.

12. The system of claim 8 wherein the imaging means includes a camera operating in the visible portion of the light spectrum.

13. The system of claim 12 wherein the memory means includes a computer memory.

14. The system of claim 13 wherein the memory means includes a disc memory.

AMENDED CLAIMS

[received by the International Bureau on 9 January 2002 (09.01.02); original claims 1-14 replaced by new claims 1-23 (5 pages)]

1. A method of rapidly reviewing recorded video images of a scene to locate those images in which a predetermined amount of activity has occurred, comprising: obtaining a plurality images of a scene over a period of time, one of the images being a reference image of the scene; comparing the reference image with other images of the scene and creating a difference image for each comparison, the difference image comprising a set of values representing a change, if any, between corresponding pixels in the respective images; performing a threshold test on the contents of the difference image to determine the number of difference image contents exceeding a predetermined threshold and counting that number; storing the images and threshold test results in a memory; and, subsequently viewing selected stored images including: setting a value representative of the amount of activity occurring in an image before it is displayed for viewing; reviewing the stored images and identifying those having threshold test results which equal or exceed the value; and, displaying the identified images, setting of the value and reviewing the threshold test results for the images allowing a large number of images to be rapidly reviewed without having to view the images in their sequential order thereby saving a substantial amount of time while allowing the user to quickly locate pertinent information related to activities at the scene.

2. The method of claim 1 wherein the identified images are displayed in their temporal sequence. 3. The method of claim 1 further including creating a separate file of the stored images identified as having threshold test results which equal or exceed the value.

4. The method of claim 3 wherein the images stored in the separate file are also stored in their temporal sequence. 5. The method of claim 1 wherein images of the scene are obtained using an imaging means including a camera operating in the visible portion of the light spectrum.

6. The method of claim 4 wherein the imaging means includes an infrared camera. 7. The method of claim 5 wherein the images obtained from the viewing means are analog signals and the method includes converting the analog signals to digital signals.

8. The method of claim 1 wherein the memory means includes a computer memory. 9. The system of claim 8 wherein the memory means further includes a disc memory.

10. The method of claim 1 wherein the identified images are displayed in reverse temporal sequence.

11. The method of claim 1 wherein the images are displayed at the rate at which they were generated.

12. The method of claim 1 wherein the images are displayed at a rate other than that at which they were generated. 13. The method of claim 1 wherein the display device contains a control to manually instruct the display device to skip to the next image which exceeds the threshold.

14. The method of claim 1 wherein the display device automatically skips to the next image which exceeds the threshold.

15. The method of claim 1 further including applying a positive mask to the image such that only the portion of the image within the mask is considered for purposes of skipping to the next image.

16. The method of claim 15 further including applying a negative mask to the image such that only the portion of the image outside the mask is considered for purposes of skipping to the next image.

17. The method of claim 1 wherein the identification of the portions of the image that have changed is stored in such a way that portions of the image may be searched independently of other portions.

18. The method of claim 3 wherein the searching is done in an automatic method without need of operator intervention.

19. A image processing system for monitoring a scene comprising: imaging means for obtaining a plurality images of the scene, one of the images being a reference image of the scene and each image obtained being represented by an analog signal; converter means converting the analog signals to digital signals; signal processor means including a comparator comparing the reference image with other images of the scene and creating a difference image for each comparison, each difference image comprising a set of values representing a change, if any, between corresponding pixels in the respective images, means performing a threshold test on the contents of the difference image to determine the number of difference image contents exceeding a predetermined threshold and counting that number, and memory means for storing the images and the threshold test results; and, viewing means by which selected stored images can be subsequently viewed, the viewing means including means for setting a value representative of the amount of activity occurring in an image, the stored images being reviewed to identify those whose threshold test results equal or exceed the value, and a monitor for displaying the identified images, setting of the value and reviewing the threshold test results for the images allowing a large number of images to be rapidly reviewed without having to view the images in their sequential order thereby saving a substantial amount of time while allowing the user to quickly locate pertinent information related to activities at the scene.

20. The system of claim 19 wherein viewing means displays the identified images in their temporal order.

21. The system of claim 19 further including creating a separate file of the stored images identified as having threshold test results which equal or exceed the value.

22. The system of claim 21 wherein the images stored in the separate file are also stored in their temporal sequence.

23. The system of claim 19 wherein the imaging means includes a camera operating in the visible portion of the light spectrum.

STATEMENT UNDER ARTICLE 19 (1)

In response to the International Search Report of October 29, 2001 , please substitute pages 7-12 with the following claims 1-23 for the pages 7-10 with the claims 1-23 currently pending in this application. Claims 1-23 are pending in the application. The language of claims 1 and 19 has been amended from the language in the claims as initially filed to better distinguish applicants' invention over the art cited in the International Search Report, particularly the Ahmad United States patent 6,009,204 (the '204 patent).

Applicants' method of claim 1, and their apparatus of claim 19, are directed to image processing in which a number of images representing a sequence of images obtained over a period of time can be processed so if it becomes necessary, or desirable, to locate something of interest occurring during the interval in which the images were obtained captured, the images can be rapidly "scanned" to locate those images containing the matter of interest. Applicants' method and apparatus save a significant amount of time since a viewer now does not have to sequentially review the images. This result is achieved by processing sequentially obtained images with a reference image. This yields a difference image. The contents of the difference image is a set of values representing the difference in pixel values for each corresponding location within the image matrix (m,n). If there is no difference in pixel values between the reference image and a subsequent image, a value of 0 is stored. Otherwise, a difference value is stored.

For subsequent rapid scanning of the images, each difference image is subjected to a threshold test to determine the variability in the image, compared with the reference image, as a function of the stored pixel value differences comprising the difference image. The images and the test results are now stored in a memory for subsequent use when location of an event or matter of interest is required. Now, the memory is quickly searched to locate those difference images whose results exceed the threshold, and the images for which those difference images were obtained are now rapidly accessed and viewed.

With respect to the cited references, and in particular the '204 patent; while the teachings of these patents relate to image processing, including the creation of difference images obtained by comparing subsequently acquired images to a reference image, Applicants' invention differs significantly from them.

The '204 patent teaches generation of a single parameter or a plurality of single parameters to describe each image obtained. Each parameter comprises a value representing the image as a whole ('204 patent, column 3, lines 45-54). These parameters are now used to find optimal reference images which are in turn used for better compression of a stream of video images. ('204 patent, column 6, lines 51-61.) Contrastingly, Applicants' invention is not concerned with finding optimal reference images. Applicants' instead choose a reference frame from every set of N number of frames, and a set number of images for the reference image. In this regard, their method is similar to prior art methods described in the '204 patent at column 7, lines 33-38, and column 8, lines 1-2. As such, Applicants' method and apparatus do not function in the way as taught in the '204 patent.

Importantly, and unlike both the '204 patent and other cited art, Applicants' do not generate single valued parameters for each image which depend only on the current image. Applicants' method instead involves generation of a parameter or parameters which relate to the difference between a current image and a previous reference image. A fundamental difference between Applicants' invention and the '204 patent is that the '204 patent teaches generating values for each image independent of any other image. Applicant's method, on the other hand, generates values based on image differences. Whereas the '204 patent uses the individual representations of each image in subsequent processing, Applicant's method uses the image difference information in subsequent processing. In addition to the above, in a preferred embodiment of their invention, Applicant's method involves creation of a table of macro block locations for each image. Included in this table is information as to whether individual pixels comprising the macro block have exceeded thresholds of intensity difference with corresponding macro block of the reference image (Specification, page 7, lines 3-10). An entire table of values is therefore kept for each image rather than a single value for the image as taught by the cited art including the '204 patent.

This should not be confused with the teachings of the '204 patent relating to allowance for use of a plurality of single parameters. This is because each of these single parameters is a measure of a different aspect of each image. ('204 patent, column 5, lines 20-22). In their invention, the contents of applicants' table represent the same aspect of the image at different locations within the image. Contrastingly, the parameters used in the '204 patent measure a single aspect of the image over the entire image.

Further with respect to the '204 patent, the patent teaches (beginning at column 5, line 37), "...calculating the difference in image values associated with video images...". As such, the '204 patent is concerned with image value differences, not differences in the images themselves. Importantly, there is no teaching or suggestion in the '204 patent of "performing a threshold test on the contents of the difference image to determine the number of difference images content", as Applicants' claim 1 requires. Rather, any threshold testing done in accordance with the teachings of the '204 patent is done is done on image value differences. Finally, the ability to rapidly review a large number of images as taught by the '204 patent refers to the fact that only the reference images are displayed for review (column 6, lines 53-61), not all of the images.

Applicants have amended their Claims 1 and 19 to clarify the distinctions between their method and apparatus and the teachings of the cited references. Applicants submit that their independent claims 1 and 19 patentably distinguish over this art and should allowed. Applicants' dependent claims 2-18, and 20-23, should also be allowable for the foregoing reasons.