中文摘要：

提取彩色图像有意义区域是目标检测和模式识别的基础。文中基于Mean Shift算法,选择合适的空间窗和色彩窗,将彩色图像聚成不同的类别,然后通过特征提取的方法将各个类别分开,最终提取出有意义区域。实验结果表明：该算法能有效地制噪声,很好地分割出感兴趣区域;与经典的Kmeans算法相比,该算法速度得到了较大的提高,分割的结果也更有意义。

英文摘要：

Color image segmentation for meaningful region is the bases of objects detection and pattern recognition. By choosing proper spatial and range kernel bandwidths, Mean Shift algorithm can divide an image into several clusters and get the meaningful region by using feature extraction. Experimental results show that, Mean Shift algorithm can effectively get rid of noise interference and acquire meaningful region. Comparing with the classical Kmeans algorithm, Mean Shift algorithm has a better segmentation results, also, the segmentation speed is greatly improved.