Digital processing techniques of time varying images are briefly reviewed. A re liable procedure for analysis of time varying images and recognition of image patterns is in detail presented. The procedure consists of the following main steps: subsequent observed images are digitized; a pre-processing is performed to reduce the noise and enhance useful patterns; pattern edges are extracted; a scanning of the pattern contours is performed; FFT of contour points is evaluated; the obtained contour FFT is compared with pre-stored data corresponding to patterns to be recognized. The good efficiency of the proposed technique isoutlined. Experimental results obtained in processing different real world varying images (in particular given by remote sensing detectors, TV or infrared cameras) are shown, confirming the good performance of the above technique for recognizing several different patterns in observed images
[1]
Jake K. Aggarwal,et al.
Computer Analysis of Planar Curvilinear Moving Images
,
1977,
IEEE Transactions on Computers.
[2]
W. B. Thompson,et al.
Combining motion and contrast for segmentation
,
1980,
IEEE Transactions on Pattern Analysis and Machine Intelligence.
[3]
H. Chien,et al.
Acquisition Of Moving Objects And Hand-Eye Coordination
,
1975,
IJCAI.
[4]
Azriel Rosenfeld,et al.
Progress in pattern recognition
,
1985
.
[5]
Thomas S. Huang,et al.
Image sequence analysis
,
1981
.
[6]
Anthony G. Constantinides,et al.
Digital filters and their applications
,
1978
.
[7]
H. H. Nagel,et al.
Image Sequence Analysis: What Can We Learn from Applications?
,
1981
.