Television – Image signal processing circuitry specific to television – Motion dependent key signal generation or scene change...
Patent
1994-12-12
1997-08-05
Chin, Tommy P.
Television
Image signal processing circuitry specific to television
Motion dependent key signal generation or scene change...
H04N 718
Patent
active
056547722
DESCRIPTION:
BRIEF SUMMARY
FIELD OF THE INVENTION
The present invention relates to a method for detecting changes in moving images, particularly to a method for detecting changes in electronically recorded moving images.
BACKGROUND INFORMATION
One field of application for detecting changes in moving images is traffic surveillance and safety engineering. By detecting changed image areas, one can automatically detect moving objects, such as vehicles in the case of an automatic traffic surveillance, or recognize unauthorized penetration in the case of object protection.
One method for detecting changes in moving images consists in segmenting an image into moving objects and a stationary background. In this method, the moving objects are treated as contiguous frames, whose movement can be described by a motion vector or a set of motion vectors.
Methods are already known, wherein the change in the video signals is compared picture element by picture element to a threshold, and the video signals are then designated as changed or unchanged. The threshold is determined according to A. Gerhard, "Motion Analysis in the Coding of Image Sequences" in the dissertation for the Communications Engineering Chair at the Munich Institute of Science and Technology Sep. 20, 1988, p. 69, in that a local value of a low-pass filtered and scaled version of a differential image, which is determined from the current image and the image that preceded in time, is subtracted from a reference threshold. The information on whether a picture element had been designated as changed or unchanged is filed in a binary mask.
However, to achieve contiguous, closed frames in accordance with this method, it is necessary to reprocess the binary mask in a further step. To this end, one applies such methods as linear filtering, erosion and dilatation, and surface-area growth.
A method for detecting moving objects is also described by D. J. Conner in "A Frames-to-blocks Picturephone Coder For Signals Containing Differential Quantizing Noise", ATT Bell System Technical Journal, vol. 52, no. 1, 1973, p. 36, et seq. This method makes use of a vertically averaged image differential signal and of an image differential signal that is compensated with respect to noise and is averaged over a surface area.
With the help of the vertically averaged image differential signal, it is first decided whether a moving edge, i.e., a moving object has been recognized, and a flip-flop is set to 1. This flip-flop is only set to 0 again, i.e., the pixel being considered is identified as a stationary object, when, in a block of 8.times.3 pixels, an accumulator counts fewer than a prescribed number of pixels whose image differential signal lies above a threshold of T.sub.2 or T.sub.3.
SUMMARY OF THE INVENTION
The method in accordance with the present invention for detecting changes in moving images has the advantage that the threshold, to which the change in the video signals is compared, is established in dependence upon the surroundings of the video signals to be decided upon. In so doing, both the temporal as well as the spatial surroundings of a block to be decided upon are considered. The threshold becomes all the lower, the more adjacent blocks have already been designated as changed when compared to their threshold. Here, the fact is taken into consideration that the probability that a block changes when its surroundings change is great, and the probability that a block changes when its surroundings do not change is small.
It is also significant that the comparison to the threshold is made on the basis of a quantity which is calculated from the picture-element variations in a plurality of pixels. Such a quantity is encumbered by substantially fewer statistical fluctuations and, thus, leads to a reliable decision.
A further advantage is that blocks of the preceding images are also used in determining the threshold. In this manner, the certainty of correctly designating a block as changed or not changed increases, since a change in a block not only depends on the adjacent blocks, but also o
REFERENCES:
patent: 4992870 (1991-02-01), Samad
patent: 4999614 (1991-03-01), Ueda et al.
patent: 5027205 (1991-06-01), Avis et al.
patent: 5091925 (1992-02-01), Haendle et al.
patent: 5103305 (1992-04-01), Watanabe
patent: 5134472 (1992-07-01), Abe
patent: 5402185 (1995-03-01), De With et al.
patent: 5404174 (1995-04-01), Sugahara
patent: 5420638 (1995-05-01), Riglet et al.
patent: 5436674 (1995-07-01), Hirabayashi et al.
J.O. Limb et al., "Combining Intraframe and Frame-to-Frame Coding for Television", The Bell System Technical Journal, vol. 53, No. 6, Jul.-Aug. 1974, pp. 1137-1173.
D.J. Connor et al., "A Frame-to-Frame Picturephone Coder for Signals Containing Differential Quantizing Noise", The Bell System Technical Journal, vol. 52, No. 1, Jan. 1973, pp. 35-51.
Carnimeo et al., "Room Monitoring Using Intraframe Adaptive Filtering Applied to Image Differences", Automazione E Strumentazione, No. 4, Apr. 1990, pp. 125-128 (Abstract only translation).
A Gerhard, "Motion Analysis in the Coding of Image Sequences", Dissertation, Munich Institute of Science and Technology, 20 Sep. 1988, pp. 66-70, 192-198 (no translation).
Aach Til
Mester Rudolf
Chin Tommy P.
Lee Richard
Robert & Bosch GmbH
LandOfFree
Method for change detection in moving images 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 for change detection in moving images, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for change detection in moving images will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-1078421