中文摘要：

针对目前基于内容的彩色图像检索方法中存在的对色彩轻微变化较敏感、运算量大、检索效果不理想等问题,提出一种新的彩色图像检索方法。首先,将HSV分量非均匀量化;然后,为了消除量化误差,将相邻量化区间对应的直方图值求和,从而构造出一种三维量化颜色直方图;针对该直方图,提出对应的三维直方图相交匹配方法来实现图像的匹配。实验结果表明该方法对彩色图像的检索效果好、运算量小、对于颜色变化鲁棒性强,可用于彩色图像的检索。

英文摘要：

To solve the problems of sensitive to slight colour change,complex in computation and poor retrieving effect that exist in current content-based colour image retrieval,a new retrieval method for colour image is proposed.First,the components of HSV are conducted the non-uniform quantisation.Then,for eliminating the quantisation error,the corresponding values of histogram in adjacent quantisation intervals are added up to form a three-dimensional colour histogram.For this histogram,a corresponding three-dimensional histogram intersection matching method is proposed to implement image matching.Experiments show that the new method is effective to colour image retrieval.It is also less complex in computation and robust to little colour change,can be used in colour image retrieval.