基于最小二乘支持向量机的图像边缘检测研究

本文研究了基于最小二乘支持向量机(LS-SVM)的图像边缘检测技术,利用LS-SVM对图像像素邻域的灰度值进行曲面拟合,通过采用多项式核函数、高斯核函数推导出图像的梯度和零交叉算子,并结合梯度算子和零交叉算子实现了图像边缘定位.通过实验获取了不同核函数的最佳卷积核的大小,同时采用遗传算法对不同核函数的参数进行寻优以获得最佳的边缘检测性能.通过与Canny方法的实验比较,验证了本文提出的边缘检测方法是有效的.

Edge Detection Based on Least Squares Support Vector Machines

A novel edge detection algorithm based on the combination results of gradient and zero crossings is presented which the image intensity of neighborhood region of pixel is well estimated by Least Squares Support Vector Machines(LS-SVM) and the gradient operator and zero crossings operator are obtained by LS-SVM based on polynomial and Gaussian kernel function.Optimal convolution kernels size is obtained by experiments when different kernel functions are applied and free parameters of kernel functions are also be optimized by genetic algorithm in order to achieve best edge detection performance.Compared with Canny method,it is shown that the proposed method based on LS-SVM is effective and fast.