基于直方图小波变换的图像颜色特征索引技术

Index of Image Color Feature Based on Wavelet Transformation of Histogram

提出了一种基于直方图小波变换的图像颜色特征索引技术:对图像HSV颜色直方图进行小波变 换,变换后数据用于相似性的递进匹配。该方法与直接直方图比较具有相同的效果,而特征矢量长度明 显减少,检索效率明显改进。实验结果表明,小波变换由于其金字塔形式可为图像数据库提供内容特征 的递进匹配,明显加快图像检索匹配速度。

The application of wavelet transformation in content-based image retrieval is discussed. The wavelet transformation is first made on the histogram of HSV color,after it is made the data are used for matching the similarity successively. Compared with direct histogram,the method has the similar effect,while the length of feature vector is reduced obviously with the effect of retrieval improved evidently. The test result shows that because of the pyramid mode of the wavelet transformation,the successive matching of content feature can be provided for the image database,with which the matching velocity is evidently expedited.