【摘要】 论文提出了一种新的基于区域的图像检索方法。与传统的基于区域的检索方法相比,论文从组成目标对象的基本结构角度出发分割图像,利用少量色彩等级更易描述对象主要构成的特性提取一组能够描述对象基本组成的区域序列。采用这些区域列的面积作为图像特征,用于图像检索。实验表明,该算法简单但非常有效,而且对图像的旋转、尺度变化具有良好的鲁棒性。更多还原

【Abstract】 This paper presents a novel region-based image retrieval approach.Compared with other traditional image segmentation approaches,it proposes a new segmentation method that reduces color level to make the parts of an object clear,so as to capture the ordered regions to describe the basic compositions of an object.This paper adopt s the area feature of these regions to retrieval image.Experiment results show that this approach is simple but effective and robust to the rotation and scaling of image contents.更多还原