Television – Image signal processing circuitry specific to television – Motion vector generation
Reexamination Certificate
1998-06-30
2002-12-10
Rao, Andy (Department: 2613)
Television
Image signal processing circuitry specific to television
Motion vector generation
C348S169000, C348S170000, C375S240170
Reexamination Certificate
active
06493041
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a method and apparatus for detecting motion in video.
2. Background Art
There are many situations in which a motion detector is used to trigger an event when motion is detected or not detected. Some applications involve turning on lights when someone enters a room, or turning off lights when there is no movement in a room. Other uses include security, car theft protection, alarms, automatic doors, and others. Current motion detection systems have a number of disadvantages, including cost, complexity, poor performance, and others.
In the prior art there are two approaches to motion detection: “active” and “passive.” Active techniques emit some form of energy (e.g. sound or electromagnetic radiation) and detect motion based on the returned signals. These techniques tend to require more power, to be more disruptive of the environment, and to be easy to detect and defeat. Passive techniques do not emit signals but instead passively observe the environment being monitored and react to observed motion. Video cameras are used in some passive motion detection techniques.
A number of techniques have been developed to detect motion within the field of view of a video camera. These techniques include analog and digital techniques. Analog techniques typically look at the analog video signal generated by a camera and detect motion by examining changes in the signal. Examples of simple prior art analog techniques include putting photocells on a television monitor and detecting changes in values, using one-shot timers to sample fixed locations in a video signal, and using various circuits to integrate the video signal. These simple techniques generate signals that can be compared against baseline values to detect changes in the video signal that presumably are caused by motion. Other prior art analog techniques filter or integrate the incoming video signal and look for gross changes in the signal's characteristics to detect motion.
These analog approaches tend to be inexpensive, but provide poor results because they utilize adulterated and simplified versions of the video signal. The bulk of the information content of the signal is discarded. Working with a signal with so little information content, the best that can be achieved is a presumption that motion has occurred in the scene when the incoming signal changes in a particular way.
All of these prior art analog techniques tend to be imprecise in what they measure. Accordingly, they have inherent limitations as to their sensitivity to actual motion. They are also susceptible to false triggers.
Digital techniques tend to be better at reducing both false positive (detecting motion when there is none) and false negative (not detection motion when motion does exist) motion detection outputs. Digital approaches are able to accurately and repeatably associate a numerical value with a physical portion of the video camera's field of view. This ability to accurately quantify the light coming from an area in space makes it possible to determine when motion occurs in the scene being observed more accurately than can be done using analog techniques.
Prior Art Digital Techniques
Digital motion detection techniques are used for two general types of applications—determining inter-video-frame motion so that signal processing can be applied to deal with video interlacing issues, and video-based monitoring for physical security purposes. Techniques developed for video interlace signal processing tend to be much more computationally intensive, and therefore costly, than techniques developed for video security monitoring. In addition, video interlace processing techniques are not suited for detecting small amounts of motion and therefore do not work well in security video applications. Because these two application areas have quite different requirements, the digital processing techniques developed for each are different in nature. For example, in the case of motion detection for the purpose of video monitoring of an area, the ability to successfully detect motion is the key objective. Exact information on which particular objects in the field of view have moved and by how much is of lesser significance. For video interlace processing, however, it is important to know which object has moved and by how much. An example of a video motion detection technique designed for: video interlace processing is disclosed in U.S. Pat. No. 4,851,904 issued to Miyazaki, et. al.
Image understanding techniques have been developed for use in video interlace processing. These image understanding techniques automatically segment a video image into regions of pixels that correspond to objects in a video camera's field of view. The motion of these objects can then be detected and tracked. These techniques are computationally intensive and expensive. They can rarely be made to run in real-time. Accordingly, they typically cannot be used for digital video security applications.
One example of a prior art computationally intensive approach for detecting motion involves taking regions of pixels (typically an N×M rectangle) from an incoming video stream and correlating them with corresponding regions of pixels in a reference image. This approach can be thought of as an approximation of the generalized image understanding approach described above. The incoming image is divided up into rectangles. These rectangles are compared against corresponding rectangles of a reference image. Dividing an image into rectangles in this manner and comparing rectangles is considerably simpler than trying to identify individual objects in an incoming image and attempting to compare the location of those objects with the location of corresponding objects in the reference image. This technique is used as part of the MPEG video compression standard and is known as “motion-compensation.” While this approach can be effective in detecting motion and is less complex than some other image understanding techniques, it is still time consuming and typically requires the use of large and expensive custom integrated circuits. In addition, it tends to be sensitive the to the quality of the incoming image. Any noise in the incoming video signal makes it very difficult to locate corresponding regions in a reference image.
Other digital techniques for motion detection in security video applications are based on the detection of edges in video images—i.e., abrupt transitions in color or brightness that delineate one region from another. Edge detection simplifies the processing of images by requiring the detection and storage of transitions only, as opposed to processing and storing values for large numbers of pixels. Edge detection takes advantage of the fact that there is a high degree of correlation between pixels in a video image (i.e., large regions of pixels tend to share similar values).
Devices that use edge detection tend to be very sensitive to false trigger events caused by changes in lighting. A stationary scene may appear to move as the lighting changes the location of shadows in a scene over the course of a day. An example of an edge detection system is disclosed in U.S. Pat. No. 4,894,716 issued to Aschwanden et al. The system disclosed by Aschwanden et al. looks for changes in the location of edges from frame to frame. This system requires a certain degree of vertical coherence to cause a trigger—i.e., there must be a given amount of phase shift of an edge across multiple lines for motion to be detected. The reference data that is stored comprises a set of counts indicating where edges exist in the vertical scan lines of the previous frame.
Edges are detected by low-pass filtering a scan line of the incoming video, thresholding the signal, then using the filtered and thresholded signal to trigger a one-shot. The one-shot in turn is used to gate a counter whose final value is the location of an edge in the scan line.
While this edge detection technique provides a simple method for motion-detection, it
Butcher Lawrence L.
Hanko James G.
Northcutt J. Duane
Wall Gerard A.
O'Melveny & Myers LLP
Rao Andy
Sun Microsystems Inc.
LandOfFree
Method and apparatus for the detection of motion in video does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for the detection of motion in video, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for the detection of motion in video will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2994372