基于矢量运算的彩色图像边缘检测算法

An Edge Detection Algorithm for Color Image Based on Vector Operation

在矢量运算基础上提出一种加权的矢量相似度彩色图像边缘检测法,给出了算法思想和计算步骤,并进行了验证。该算法首先将彩色图像转换到YCbCr颜色模型并构造像素的矢量表达,然后利用矢量欧氏距离计算像素间的相似度分布,相似度具有局域最小特征的点即为边缘。该算法通过颜色模型转换实现亮度和颜色分离,从而可以引入权值控制亮度和颜色对边缘的贡献。实验结果表明,矢量空间相似度可以很好地表达图像边缘,亮度和颜色分离的方式也有利于彩色图像的边缘检测。

Based on vector operation, an edge detection algorithm of a weighted vector similarity is presented in this paper. Both principles and algorithm are all discussed in this paper. The verification is given. The color image is transformed into YCbCr model and is expressed by vector for each pixel. The euclidean distance of vectors is engaged to calculate the similarity between pixels. The pixels with local minimum value of similarity are considered as points on edge. The algorithm can easily control the extent of contribution be-tween luminance and color by adjusting the weight in edge detection. The result shows that the similarity of vectors is valid for edge ex-pression and that separating luminance and color is beneficial for edge detection.