基于逼近增强算子的合成孔径雷达图像特征提取算法  

Synthetic aperture radar image feature extraction algorithms based on approximation enhancement operator

作　　者：王平[1] 纪钢[2] 

机构地区：[1]重庆医科大学附属第一医院,重庆400016 [2]重庆理工大学,重庆400050

出　　处：《计算机应用研究》2015年第7期2238-2240,共3页Application Research of Computers

基　　金：国防科工委资助项目

摘　　要：针对含噪SAR图像的边缘检测效果不理想、边缘特征不明显等问题,提出一种基于逼近增强算子的合成孔径雷达(SAR)图像特征提取算法。该算法利用多尺度非均匀滤波将含有噪声与不含噪声的像素点的灰度值、结构元素以及区域内的像素加权灰度密度这三个特征进行区分,以达到去噪效果。采用基于增强算子的SAR图像检测方法,通过SAR图像的像素灰度值以及像素点分布密度均值来计算综合均值阈值,通过阈值来判断像素点是否属于边缘部分。在实验中,通过分别与基于修改的LSD算法、基于水平集算法以及基于核心聚类算法的SAR图像提取方法进行了对比分析,从对比结果可以得出算法在对含噪SAR图像进行边缘检测时可以得到更明显的边缘信息。For edge detection and obvious edge features is not ideal and other issues in noisy SAR images, this paper proposed synthetic aperture radar image feature extraction algorithms based on approximation enhancement operator to enhance operator, which used the multi-scale non-uniform filter containing gray value noise and noise-free pixels, structural elements, and pixel density weighted gray area of these three characteristics distinguish ,to achieve the effect of denoising. Using SAR image detection method based on the use of enhanced operator to calculate the average integrated density threshold distribution and the mean gray value of the pixel of the pixels of the image by a threshold value to determine whether the pixel belonged to an edge portion. In the experiment, by comparative analysing with SAR image extraction method based on LSD modified algorithm and level set algorithm and clustering algorithm, the results can be drawn from the comparison, this algorithm in noisy SAR image edge detection can be more obvious edge information.

关 键 词：SAR图像检测 特征提取 多尺度非均匀滤波 逼近增强算子 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]