Translator Disclaimer
22 October 2010 Texture recognition of SAR image based on surfacelet transform
Author Affiliations +
In order to take full use of SAR image dark areas information, texture recognition method of SAR image based on Surfacelet transform is proposed. Firstly, SAR image in test samples is decomposed into brightness SAR image and reflection SAR image based on Retinex model; secondly, the decomposed SAR image carries through Surfacelet transform, in order to reduce computational complexity and shorten the follow-up recognition time, the decomposed SAR image is processed two dimension three level scale Surfacelet transform, and the first level is 8 directions, the others are 8,2 directions respectively, after image Surfacelet transform, the sparsity of the distribution coefficient is even more evident, especially high-frequency sub-band; thirdly, coefficient matrixes of high-frequency sub-band reduce dimension processing by using the singular value decomposition theory of matrix; fourthly, the final eigenvector are composed of character of energy of coefficient matrix in low-frequency sub-band and first and second-order moment of coefficient matrix in intermediate frequency sub-band and eigenvector by reducing dimension in high-frequency subband; at last, use BP neural network to train with supervised and recognize of samples. The method takes full use of SAR image dark areas information, and the simulation results show that the method the recognition rate is better than other methods.
© (2010) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Dandan Liu, Can Zhao, and Chunrui Tang "Texture recognition of SAR image based on surfacelet transform", Proc. SPIE 7658, 5th International Symposium on Advanced Optical Manufacturing and Testing Technologies: Optoelectronic Materials and Devices for Detector, Imager, Display, and Energy Conversion Technology, 76583R (22 October 2010);

Back to Top