This paper addresses the problem of detecting small, moving, low amplitude in image sequences that also contain moving nuisance objects and background noise. Rough sets (RS) theory is applied in similarity relation instead of equivalence relation to solve clustering issue. We propose fractal-based texture analysis to describe texture coarseness and locally adaptive threshold technique to seek latent object point. Finally, according to temporal and spatial correlations between different frames, the singular points can be filtered. We demonstrate the effectiveness of the technique by applying it to real infrared image sequences containing targets of opportunity and evolving cloud clutter. The experimental results show that the algorithm can effectively increase detection probability and has robustness.
[1]
Janusz Zalewski,et al.
Rough sets: Theoretical aspects of reasoning about data
,
1996
.
[2]
Andrzej Skowron,et al.
The Discernibility Matrices and Functions in Information Systems
,
1992,
Intelligent Decision Support.
[3]
Dimitrios Charalampidis.
Novel textural features and techniques for image segmentation and classification
,
2001
.
[4]
Krzysztof Krawiec,et al.
ROUGH SET REDUCTION OF ATTRIBUTES AND THEIR DOMAINS FOR NEURAL NETWORKS
,
1995,
Comput. Intell..
[5]
C. A. Murthy,et al.
Thresholding in edge detection: a statistical approach
,
2004,
IEEE Transactions on Image Processing.