模糊集理论在彩色图像检索中的应用

Application of fuzzy set theory to color image retrieval

把模糊集理论中的模糊关系应用于对色彩相似性的判断 ,并在彩色图像的检索中进行了应用 .其中仅使用了一种基于色彩空间的矢量表达的隶属函数即可分别计算出待检索图像与库图像对应特征色之间的相似性程度 ,并用α级关系来抽取其中的一些彩色图像 ,给出检索结果 .在进行色彩相似性的计算中还采用了逐级筛选法及色彩矢量分区比较法 ,从而避免了大量涉及无关的色彩矢量的计算 ,有效地降低了红绿蓝 (RGB)空间的色彩相关性 。

We report an application of fuzzy relations in fuzzy set theory to the calculation of the similarity between colors, and apply it to the color image retrieval, where only a membership function is used as a measure of-the similarity between the characteristic color vectors of a query. image and of the images in the database. A match between two colors is,defined as the alpha-cut fuzzy relation. In the comparison process, screening with decreasing by orders for comparisons and a method of regional color vector comparison are developed, which effectively reduce the color correlation in the red-green-blue (RGB) space and save from massive unnecessary calculations. Our theoretical and experimental results show clearly the advantages of our approach; in both accuracy and speed of the color image retrieval.