交互信息理论及改进的颜色量化方法在图像检索中的应用研究

Research to the Application of Mutual Information and Improved Color Quantization Method in Content Based Image Retrieval

设计实现了基于内容的图像检索原型系统与检索性能评价系统.提出了一种基于交互信息与信息熵的图像相似性度量方法—交互信息距离(M ID)和一种改进的颜色量化方法,成功地将两种方法应用于基于颜色特征的图像检索中,通过比较性研究证明,M ID能较KLD提供更高的检索准确率;改进的颜色量化方法较基于HSV颜色空间的一致颜色量化方法有效地提高了检索准确率.试验证明,颜色空间的合理量化对图像检索有着重要影响,在选择颜色空间进行图像检索的同时,不能忽略对颜色空间的合理量化.

This paper designs a content based image retrieval（CBIR） system and a retrieval performance evaluation system, and proposed a similarity measurement-mutual information distance, which is based on the mutual information and Shannon entropy. The approach is hased on the premise that two similar images should have high mutual information, or equivalently, the querying image should convey high information about those similar to it. The method is successfully applied in color based image retrieval system. A comparative study shows that the MID is a more effective image similarity measure than Kullback- Leibler distance. An improved color quantization method is proposed. Experimental results demonstrate that the reasonable quantization to color space is important to the retrieval effectiveness. When choosing color space for image retrieval we can＇t neglect the importance of color quantization.