This paper presents the configuration of an image processing system for efficiently detecting a moving target from infrared images of a scene with a complicated background such as an urban area, and improving the target detection accuracy by optimizing the weights of the feature vectors by genetic algorithms (GA). The proposed system applies edge-enhancing binary processing with a background suppressing function and motion vector processing with a moving body detection function in parallel to the input infrared images. It performs target decision or detection processing by using eight features obtained by the above two kinds of processing. In order to improve the identification capability of target decision or detection, optimization of the weights of the feature vectors is performed by means of a GA. In addition, an effort is made to improve target detection accuracy by using gate processing to limit the processed area. The effectiveness of the proposed system is verified by applying it to the detection of an airplane against the background of an urban area. The proposed system attempts to shorten the processing time by parallel processing by a simplified configuration that combines the basic processes, assuming its application to guided mechanism with space and processing time constraints, and may be applied to traffic monitoring, harbor monitoring, field search monitoring systems, and the like. © 2003 Wiley Periodicals, Inc. Syst Comp Jpn, 34(7): 76–86, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.10168
[1]
E. D. Evans,et al.
Search Radar Detection and Track with the Hough Transform
,
1994
.
[2]
Robin J. Evans,et al.
Peformance of dynamic programming techniques for Track-Before-Detect
,
1996
.
[3]
S. M. Tonissen,et al.
Maximum likelihood track-before-detect with fluctuating target amplitude
,
1998
.
[4]
Michael C. Wicks,et al.
A statistical theory for optimal detection of moving objects in variable corruptive noise
,
1999,
IEEE Trans. Image Process..
[5]
Dana H. Brooks,et al.
Point target detection in IR image sequences based on target and clutter temporal profile modeling
,
1999,
Defense, Security, and Sensing.
[6]
Shigenobu Kobayashi,et al.
Recognition process using feature data fusion for imaging system
,
1999,
IEEE SMC'99 Conference Proceedings. 1999 IEEE International Conference on Systems, Man, and Cybernetics (Cat. No.99CH37028).
[7]
Shigenobu Kobayashi,et al.
Infrared image target detection process using continuous features
,
2000,
SPIE Defense + Commercial Sensing.