图像边缘检测技术的改进  

Improving of image edge detection technique

作　　者：厉丹[1,2] 钱建生[1] 芦楠楠[1] 周均民 田文科 

机构地区：[1]中国矿业大学信息与电气工程学院,江苏徐州221008 [2]中国矿业大学徐海学院,江苏徐州221008 [3]鲁能菏泽煤电开发有限公司,山东菏泽274000 [4]潞安环能股份公司王庄煤矿,山西潞安046204

出　　处：《计算机工程与应用》2010年第18期164-166,共3页Computer Engineering and Applications

基　　金：国家自然科学基金 No.70533050;the National Natural Science Foundation of China under Grant No.70533050

摘　　要：比较了Roberts算子、Sobel算子、Gauss-Laplace算子、Canny算子等几种传统的边缘检测技术,指出这些算法在边缘检测精度和抗噪声性能方面存在一定的问题。在分析以上算子的缺陷后提出改进的方法。通过迭代算法寻找最佳阈值,增强了目标和背景的对比和目标边缘,准确提取目标区域,并结合形态学进行轮廓提取,有效减少噪声对灰度门限值的影响,通过对实验图像的分析表明,改进的检测算法对图像边缘提取具有较好的检测精度,抗噪能力和准确性。Compared with some traditional edge detection technique such as Roberts operator,Sobel operator,Gauss-Laplace operator,Canny operator,etc.This paper indicates the traditional algorithms that have drawbacks in the detection accuracy and the antinoise performance.After analyzing the limitation of the above operators,a new method is put forward.On researching for best boundary through iteration arithmetic,it enhances contrast of object and background,object edge,and abstracted object area exactly.It reduces influence of noise to gray threshold in combined with morphology.The results of the experiment pictures indicate that the improved algorithm has the better accuracy,noise against quality,and precision in the edge detection.

关 键 词：图像处理 边缘检测 迭代 腐蚀 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]