基于IIR指数平滑滤波与非线性Laplacian算子的边缘检测方法  

AN EDGE DETECTION APPROACH VIA IIR EXPONENTIAL SMOOTHING FILTERING AND NONLINEAR LAPLACIAN OPERATOR

作　　者：游素亚[1] 柳健 徐光佑[1] 

机构地区：[1]清华大学计算机科学与技术系

出　　处：《计算机学报》1997年第7期667-672,共6页Chinese Journal of Computers

基　　金：国家自然科学基金;测绘遥感信息国家重点实验室基金

摘　　要：本文提出了一种基于ＩＩＲ指数平滑滤波与非线性Ｌａｐｌａｃｉａｎ算子的边缘检测方法，具有以下特点：采用最佳指数平滑滤波器增强抗噪声能力，且可实现无限大窗口的快速递归运算；利用非线性Ｌａｐｌａｃｉａｎ算子只沿边缘方向增强边缘，充分利用了边缘的局部方向信息；引入边缘强度信息，在二阶零交叉处根据边缘强度定位边缘；运算量独立于滤波器窗口尺寸，具有较高的计算效率．通过与ＬＯＧ算子的对比实验和在立体视觉中的应用表明，该方法具有良好的实用性．An edge detection approach via IIR exponential smoothing filtering and nonlinear Laplacian operator is proposed, which has the following advantages. At first, utilizing the exponential kind filter can greatly enhance the ability of smoothing noise, and can implement fast structure recurred filter. Secondly, the nonlinear Laplacian operator can adapt the edge local information and enhance edges only along the edge directions. Thirdly, the edges are located by combination edge strong information with the zero crossings. Finally, the operation which is independent of the size of the filter is efficient. The experimental results of comparing the proposed approach with the LOG operator and the application in stereo vision show that the proposed approach has greatly practical value.

关 键 词：边缘检测 IIR指数滤波器 非线性 LAPLACIAN算子 

分 类 号：TP301[自动化与计算机技术—计算机系统结构] TP391.4[自动化与计算机技术—计算机科学与技术]