基于量化颜色空间的彩色图像检索算法

Method of Color Image Retrieval Based on Quantified Color Space

研究符合人的视觉特性的HSV颜色空间中,对彩色图像进行量化和再次量化处理,克服了颜色的连续性对图像检索带来的影响,将第一次量化和经过再次量化得到的结果作为图像特征。针对欧式距离在检索过程中没有考虑到色彩之间相似性的局限性,为改善图像检索性能,在进行相似匹配时引入了一个色彩相似矩阵,利用直方图二次距离作为相似度量公式对彩色图像库进行检索,结果表明,利用经过再次量化得到的图像特征进行检索比仅一次量化之后得到的图像特征进行检索具有很高的鲁棒性。

In HSV color space that is the best visual identity of people,color images can be quantified and quantitative handle again,so the impact of the continuity of the color is overcome in retrieval.the Euclidean distance is used in the retrieval of a Color image database.Because the Euclidean distance in retrieval process does not take into account similarities between the color limitations,and so the introduction of a similar color,matching similar matrix,using histogram quadratic distance formula as a similar measure to achieve a more satisfactory results.