Research Of Color Image Segmentation Technology

Separating valuable blocks from images is a fundamental part of machine vision and pattern recognition technology.Because color images have more content,so it contains more information than gray image,with the diversification of color image,the presentation of complex,color images by people more and more applied to various occasions.This paper proposes a new marker watershed segmentation algorithm based on HSV color space model,based on maximum gradient of multi-scale structural elements of the mark watershed transformation and growth algorithm combined with automatic seed blocks,specific content is as follows:First of all,the original color image from RGB color model transformation for HSV color model,in order for H,S,V three components,selection of maximum radius to m "disk" structure elements,the elements of the size structure based on maximum gradient realize morphological opening and closing reconstruction,and marked the deep local minimum value of the gradient image,and watershed transform.The improved watershed method combines the influence of multidimensional structure elements on the image,which can improve the segmentation phenomenon.By automatic threshold choice seeds block to block,this paper select method 1s a kind of based on color difference of the average method to select seed blocks,if the seeds of the currently selected block adjacent blocks the relative Euclidean distance of not more than a threshold T,implementation,and these two blocks and then.This not only reduces the amount of computation,but also achieves better results.Finally,a small block is merged.In this paper,the method of vc++6.0+ matlab is used to simulate the simulation,and the results show that the method is better.In order to accurately evaluate the effectiveness of the image segmentation,the 150 sample image segmentation experiments,the results show that the segmentation success rate was 92%,shows that this algorithm is effective and feasible.