Unsupervised Multiresolution Image Segmentation Integrating Color and Texture

• 作者: TANGXiaofang XINGQiang YUANBaozong
• 来源期刊: 电子学报(英文版)
• 学科: 工学
• 关键词: CBIR 图像分割 多分辨图像分析 色矩 波形变换
• 年，卷（期）: 2004,（1）
• 页码范围: 49-53
• 页数: 5页
• 分类号: TP391.41
• 语种: 中文

摘要

Unsupervised segmentation of images is highly useful in various applications including contentbased image retrieval. A novel multiresolution image segmentation algorithm, designed to separate a focused object of interest from background automatically, is described in this paper. According to the principle of human vision system, our algorithm first searches the salient block representing object in global image domain. Then all image blocks are clustered using the feature of color moments and texture in salient block. At last the algorithm classifies the image blocks belonging to object class in high resolution. Experiment shows that our algorithm achieves better segmentation results at higher speed compared with the traditional image segmentation approach using global optimization.