基于SVM分类的图像边缘检测研究

Study on image edge detection based on classification using SVM

针对经典边缘检测算子检测含噪声图像边缘结果不理想的问题,提出采用基于统计学习理论的支持向量机(Support Vector Machine,SVM)方法建立分类模型,并对图像进行边缘检测.实验基于MATLAB平台,运用OSU_SVM3.0和LS_SVM工具箱建立模型,实验表明:该方法不仅可以有效地检测图像边缘,还在一定程度上克服了噪声干扰,并可以将被测图像中的特定图形检测出来.

According to the problem of bad edge of the image with noise detected using the classical edge detectors,a method based on support vector machine（SVM） of statistical learning theory is proposed to create classification model and detect the image edge.The experiment is based on MATLAB and OSU_SVM3.0 and LS_SVM toolbox are used to create the model.The experiment shows that the method can detect edges of the image very well,and overcomes the noise disturbance in some degree.The specific graphic image can be detected using the method.