Color Image Segmentation Based on Improved ISODATA
基于改进的ISODATA算法彩色图像分割

How to fast, accurately and effectively segment objects in the color images is the key point in the computer vision and image analysis. This paper introduces a method of region-based color image segmentation. This method first extracts color, texture, and location features for each pixel form integrative feature vectors by selecting suitable color space to form the feature space. In the feature space, the initial cluster center and the number are determined by the improved ISODATA algorithm adaptively, then an image is clustering and separated into regions. Finally, the features of regions are extracted. The experimental results and the comparision with the similar approach are provided. Experimental results show the proposed method has high segmentation speed and good sementation results, and it is fit for region-based image retrieval system and has better application values.