一种基于DCT系数的彩色图像检索方法

The Color Image Retrieval Method Based on DCT

基于内容的图像检索是当前图像数据库领域中的一个研究热点。给出一种直接在频域内对彩色图像进行检索的方法:将一幅真彩图像变换到YCrCb空间,提取主颜色,对主颜色进行DCT变换,变换后的低频分量即为图像的颜色表示。在此基础上,讨论了图像的相似度量以及相应的图像检索技术,并给出了实验结果和图像检索性能的评价。

The images retrieval based on the content is a hot point of the research of the current image database field. This paper puts forward a color image retrieval method used directly in the frequency-domain： a real color mages in RGB color space is transformed into YCrCb color space, dominant color is computed from the images, which is transformed by using the DCT, and the transformed low-frequency components are the color indication of the image. And based on this, this paper discusses on the similarity measuring and corresponding image retrieval techniques, and gives the evaluation on the test results and the image retrieval oerformance.