Fractal-based infrared target detection

• Published in: 2009 International Conference on Machine Learning and Cybernetics Vol. 4; pp. 2443 - 2447
• Main Authors: Yu-Qiu Sun, Xiao-Qiang Feng, Ling Li, Jin-Wen Tian, Jian Liu
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.07.2009
• Subjects: Clutter; dimension; fractal; Fractals; Infrared detectors; infrared images; Infrared imaging; Interference; Mathematical model; Noise robustness; Object detection; Pixel; Sea surface; Target detection
• ISBN: 9781424437023; 1424437024
• ISSN: 2160-133X
• DOI: 10.1109/ICMLC.2009.5212162

In this paper, we present a method for infrared image small targets detection with complicated background clutters. Fractals are mathematical models for very irregular and very detailed characterized by the fractal dimension. A fractal dimension image is formed after the fractal dimension of every pixels in an image is calculated. According to the characteristics of natural objects and man-made objects, their fractal dimension is different. The smoother the surface of an object appears, the less fractal dimension the pixel is. In this paper, the targets are military objects and the background of infrared image is sky and sea with complicated clutters. On the other hand, the infrared images have serious noise. However, targets can be detected because of the fractal dimension difference between natural surface and man-made surface. Whether or not targets are interference, the detection result is efficient. This approach results in an effective and robust technique for clutter suppression, and produces consistent detection performance over a wide range of radiant conditions. Moreover, this method provides sufficiently good target segmentation as well.