【摘要】 在图像检索中,为了在特征匹配时获取一个完整的且包括目标对象的目标区域,并减少背景及背景中其他物体对目标对象匹配的影响,根据用户需要查找的目标对象所具有的所有颜色,采用了多颜色值聚类分割的方法,提取被检索图中的目标区域,使得提取出的目标区域包括了目标对象的所有颜色,同时更好地排除了背景以及背景中其他颜色所组成的其他物体对相似性匹配的影响。试验结果表明,该方法能够准确地得到一个与目标对象相符合的目标区域,而忽略其他颜色物体对象。更多还原

【Abstract】 In image retrieval,in order to get an integral target object region and reduce background and other objects’ influence to the match,according to the colors of the needed object,multi-color cluster partition was used,and the target region in the searched image was extracted,making the target region contain all the needed object colors.The influence of the background and other color objects was reduced.The test results indicate that,the method can exactly get a target region matched with the needed object,and neglect other color objects.更多还原