基于主色提取和主色集扩充的图像检索算法

IMAGE RETRIEVAL ALGORITHM BASED ON MAIN COLOR EXTRACTION AND MAIN COLOR SET EXPANSION

提出一种基于主色提取和主色集扩充的图像检索算法。算法首先对图像进行主色特征提取,采用提出的改进主色集扩充方法分别对两图像进行主色集扩展,使其维数相同;然后,将经过主色提取得到的主色集从传统RGB颜色空间转换为具有良好视觉一致性HSV颜色空间;最后,采用Quadratic距离进行图像的相似性度量。实验证明算法易于实现、具有较好的时间复杂度及较高的查准率和查全率。

The authors put forward an image retrieval algorithm based on main color extraction and main color set expansion.At the beginning,the algorithm extracts main color feature from images;then it uses an improved main color set expansion method,which is proposed in the article,to expand main color sets of two images one by one,until they are of the same dimension.Next,it transforms the main color sets,which are acquired by main color extraction,from the traditional RGB color space to HSV color space with good visual consistency.Finally,it uses Quadratic distance to measure image similarity.Experiments have shown that the algorithm is easy to implement while it bears better time complexity,higher precision rate and recall rate.