微型机与应用
微型機與應用
미형궤여응용
MICROCOMPUTER & ITS APPLICATIONS
2014年
7期
35-37
,共3页
边缘检测%边缘细化%改进 Sobel%LoG 算子%边缘的边缘
邊緣檢測%邊緣細化%改進 Sobel%LoG 算子%邊緣的邊緣
변연검측%변연세화%개진 Sobel%LoG 산자%변연적변연
edge detection%edge thinning%improved Sobel%LoG operator%edge of edge
鉴于Sobel 算子检测边缘较粗、定位不准确,以及LoG 算子具有各向同性的特点且对边缘方向性信息检测不敏感,提出了Sobel算子与LoG算子相结合的边缘检测与细化方法。首先用水平、垂直、两个斜对角4个方向模板改进原Sobel 算子两个方向模板,并用改进的Sobel 算子对原图像进行边缘检测,得到粗边缘图像;然后使用 LoG 算子检测粗边缘图像的边缘;最后将两次边缘检测结果相减,达到边缘细化的目的。实验结果表明,该方法有效地解决了原 Sobel 算子检测边缘较粗的问题,得到的边缘较细;克服了 LoG 算子对方向性边缘信息不敏感的缺陷;且运算速度与传统 LoG 方法基本相当。
鑒于Sobel 算子檢測邊緣較粗、定位不準確,以及LoG 算子具有各嚮同性的特點且對邊緣方嚮性信息檢測不敏感,提齣瞭Sobel算子與LoG算子相結閤的邊緣檢測與細化方法。首先用水平、垂直、兩箇斜對角4箇方嚮模闆改進原Sobel 算子兩箇方嚮模闆,併用改進的Sobel 算子對原圖像進行邊緣檢測,得到粗邊緣圖像;然後使用 LoG 算子檢測粗邊緣圖像的邊緣;最後將兩次邊緣檢測結果相減,達到邊緣細化的目的。實驗結果錶明,該方法有效地解決瞭原 Sobel 算子檢測邊緣較粗的問題,得到的邊緣較細;剋服瞭 LoG 算子對方嚮性邊緣信息不敏感的缺陷;且運算速度與傳統 LoG 方法基本相噹。
감우Sobel 산자검측변연교조、정위불준학,이급LoG 산자구유각향동성적특점차대변연방향성신식검측불민감,제출료Sobel산자여LoG산자상결합적변연검측여세화방법。수선용수평、수직、량개사대각4개방향모판개진원Sobel 산자량개방향모판,병용개진적Sobel 산자대원도상진행변연검측,득도조변연도상;연후사용 LoG 산자검측조변연도상적변연;최후장량차변연검측결과상감,체도변연세화적목적。실험결과표명,해방법유효지해결료원 Sobel 산자검측변연교조적문제,득도적변연교세;극복료 LoG 산자대방향성변연신식불민감적결함;차운산속도여전통 LoG 방법기본상당。
The edges detected by Sobel operator are rough and had the poor accuracy and because of its isotropy , traditional Laplacian of Gaussian (LoG) operator is not used in many occasions. So a method for edge detection and thinning combining Sobel operator and LoG operator is put forward. The improved Sobel operator is got with four gradient templates in the horizontal , vertical and two diagonal directions and the rough edge image is got after edge detection using the improved Sobel operator. And then LoG operator is used to detect the rough edge to get a new edge. Finally, edge thinning is got with the difference between the rough edge image and the new edge image. The experimental results show that this method can thin the rough edge detected by Sobel operator effectively, has solved the defect that LoG operator has poor ability to get directional edge information and its computing speed is roughly equal to LoG operator.