基于独立成分分析的强噪声海域SAR图点目标提取方法

ICA-Based Point Target Extraction for Sea-SAR Images with Intensive Noises

将分形理论和独立成分分析(ICA,Independent Component Analysis)相结合,用于强噪声海域合成孔径雷达(SAR,Synthetic Aperture Radar)图像的点目标提取.首先依据分形理论,计算点态H lder指数,再对指数图进行二值模糊处理,接着将此图参与ICA计算,然后使用泛化可调算子收缩并优化其独立成分,实现强噪声的有效抑制,进而提取点目标.实验结果表明,H-ICA与基于ICA降噪的传统算法相比,能够有效地降低海域强噪声,并成功实现点目标提取.

Fractal and ICA （Independent Component Analysis） SAR （Synthetic Aperture Radar） images with intensive noises. are combined to detect point targets in sea - First, pointwise Holder exponent of original image is computed. Then, binary - fuzzy processing is employed to modify the Holder image and the new H - image is used as input for ICA algorithm. A generalized adjustable operator is used thereafter to enhance the independent components computed from ICA, and the point target can be detected in the recovery image finally. It is shown through the results that the proposed algorithm sharply reduces the speckle and that a successful target extraction immune to intensive noise is obtained.