A Novel Method for Point Target Detection in Sea-clutter SAR Image with Fractal and ICA
Yue Fei Zhou · Journal of Astronautics · 2007
This paper presents a novel approach for point target detection of sea-clutter SAR images.Traditional methods for this application can be classified into two aspects: threshold segmentation based on intensity difference and target extraction based on suitable denoising.However,they appear to be not effective enough especially when sea clutter is strong.Taking advantage of the essentials of both methods,a effective approach using space separation is developed based on fractal theory and independent component analysis.First,pointwise Hlder exponent are computed and binary-fuzzy processing is used for enhancement;then,basis images and independent components of the processed image are respectively obtained by ICA technique.After that,according to separation criterion,the original space is separated into two subspaces called clean-space and noise-space with respective independent components and corresponding basis images.Finally,the recovery image is obtained after enhancing the independent components in clean-space.As the results show,the proposed method is validated and point target is extracted more efficiently compared with conventional ones.