The Research On Edge Detection Algorithm Of High Resolution For Color Images

As the low-level operation in image processing technology,edge detection plays a vital role in the field of machine vision and image processing.It can lay the foundation for subsequent image processing.While,color image contains more color information which has much richer messages to describe object features.Therefore,color edge detection gains much attention of researchers.The thesis is written on the basis of monochromatic-based color edge detection techniques.It analyzes the principle of output fusion method and multidimensional gradient method,and the experimental simulations of the existing algorithms are analyzed,then experimental results are given respectively.Main research content of the thesis concentrates on multidimensional gradient method.It is aimed at studying edge which is caused by color change to get more edges.Two new color edge detection algorithms are proposed in this paper: 1)Edge detection algorithm of combining color Canny operator with anisotropic Gaussian kernels(ANGKs).The algorithm combined advantages and disadvantages of monochromatic-based techniques and vector-values techniques.Firstly,the Jacobin matrix is proposed to compute eigenvector corresponding to maximum eigenvalue as gradient by using color Canny operator,then gradient of each color channel is calculated by using ANGKs.At the same time a self-adaptive gradients fusion method is proposed.Finally color edge map can be obtained via using non-maximum suppression and double threshold processing.2)Anisotropic Gaussian kernels edge detection algorithm based on chromatic difference(ANGKs).Chromatic difference among color channels is analyzed in this method,and changes are considered that result from chromatic difference.The gradients of chromatic difference and gray image are merged together to detect color edge by using ANGKs.Finally performance of the proposed two operators is evaluated qualitatively and quantitatively for non-noise and noise images,which is compared with other operators respectively.The experimental results indicate that edge detection algorithm of combining color Canny operator with anisotropic Gaussian kernels preserves images structure information and gets more edges,while it can suppress Gaussian white noise better.Anisotropic Gaussian kernels edge detection algorithm based on chromatic difference studies on chromatic difference and gets more edges.At the same time it can get clearer image contour,and it can resist Gaussian white noise with bigger variance and gets more complete edge maps.