基于支持向量机的彩色图像边缘检测  

COLOR IMAGE EDGE DETECTION BASED ON SUPPORT VECTOR MACHINE

作　　者：左森[1] 郭晓松[1] 万敬[2] 郭君斌[1] 

机构地区：[1]第二炮兵工程学院,陕西西安710025 [2]第二炮兵装备研究院,北京10085

出　　处：《计算机应用与软件》2007年第9期151-152,161,共3页Computer Applications and Software

摘　　要：提出了一种新的基于支持向量机的彩色图像边缘检测算法。将彩色图像像素3×3邻域内像素的RGB值表示为一个27维的向量,作为该像素的特征,利用支持向量机直接判断其是否为边缘点。针对实际图像的边缘检测实验表明,支持向量机可以有效地进行彩色图像的边缘检测,其检测效果可以和传统的Sobel等边缘检测算子相当。A new method for color image edge detection based on Support Vector Machine（SVM） is presented. The RGB value of each pixel in 3×3 neighborhood of a color image is represented as a 27 components vector. The pixel can be classified as a part of an edge or not by an SVM classifier directly. The results of real images edge detection experiments show that this method is comparable to the classical edge detection methods, such as Sobel edge detector.

关 键 词：支持向量机 彩色图像 边缘检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]