基于FPGA的SAR图像快速纹理特征提取方法

Fast extraction method of SAR images texture feature based on FPGA

针对灰度共生矩阵算法及其改进算法计算量大、消耗时间长等问题,本文提出了一种基于FPGA(Field Programmable Gate Array)的SAR图像快速纹理特征提取方法.为了减小边界效应,该算法将SAR图像不均等地分成若干个小的图像块,各个图像块分别经过一个GLCM算法处理单元进行同步计算,同时通过图像纹理特征提取方法的数据重用性来进一步提高算法的处理速度.实验表明,这种并行处理方法能够大幅度地减少SAR图像特征提取时间,这为大尺寸的SAR图像的纹理特征快速提取提供了一种有效的途径.

A fast method based on FPGA is proposed to achieve texture feature extraction of SAR images, against large amount of calculation, long time consumption and so on of GLCM algorithm and its improved algorithms. In order to reduce the boundary effect, in this paper, the SAR images are unequally divided into several small image blocks, each small image block goes respectively through a GLCM algorithm processing unit for synchronous calculation. In addition, the processing speed of the algorithm is further improved by data reusability of the image texture feature extraction method. And the experi- ment results show that this parallel processing method can greatly reduce the feature extraction time of SAR images, which provides an efficient way for the rapid texture feature extraction of large-scaled SAR images.