激光主动成像图像边缘检测算法研究  

Image Edge Detection Algorithm for Laser Active Imaging

作　　者：宋睿[1] 张合新[1] 吴玉彬[1] 宫梓丰 

机构地区：[1]中国人民解放军火箭军工程大学控制工程系,陕西西安710025

出　　处：《激光与光电子学进展》2017年第8期153-160,共8页Laser & Optoelectronics Progress

基　　金：国家自然科学基金(61203189)

摘　　要：针对激光主动成像图像特点,提出一种分数阶微分与Sobel算子相结合的边缘检测算法。构造小波软阈值法与非局部均值滤波相结合的去噪算法对图像进行预处理;分析传统Sobel算子的不足,并利用分数阶微分对其进行改进,提出基于分数阶微分和Sobel算子的四方向边缘检测模型对图像进行梯度运算;运用最大类间差分法自适应选取阈值实现二值化完成边缘检测。实验结果表明,与传统边缘检测算法相比,该算法能够检测出更多的图像边缘细节,且具有更好的可匹配性参数。According to the features of laser active imaging, a new edge detection algorithm combining fractional differential and Sobel operator is proposed. The image preprocessing is completed by constructing a denoising algorithm based on the combination of wavelet soft threshold denoising and non-local means filtering. The deficiency of traditional Sobel operator is analyzed. Then the Sobel edge detection operator is improved by fractional differential, and a four-direction edge detection model based on the fractional differential and Sobel operator is proposed to complete image gradient operation. The maximum interclass differential method is used to automatically select the threshold, so as to achieve the binarization to complete edge detection. Experimental results show that compared with the traditional edge detection algorithms, the proposed algorithm can detect more image edge details and has better matching parameters.

关 键 词：图像处理 激光主动成像 边缘检测 SOBEL算子 分数阶微分 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TN958.98[自动化与计算机技术—计算机科学与技术]