随着图像数据的海量增长,图像检索效率逐渐成为研究热点。为了提高图像检索的准确率,提出了一种改进的HSV颜色空间量化方法,细化色调H的分类,使量化结果更接近人类感知,并在此基础上采用分块策略进行仿真实验。实验结果表明,提出的方法能更好地描述图像的颜色特征,效果令人满意,具有一定的实用性。
With the massive growth of images,the efficiency of image retrieval gradually becomes a research hotspot. In order to improve accuracy of the image retrieval,Hue-Saturation-Value( HSV) color space is improved,which details the classification of H hue and makes the quantitative results more close to human perception. On this basis, block color histogram is extracted as a retrieval feature. Experiments show that the proposed method can better describe color feature of images,the results are satisfactory and have some practical value.