基于颜色和轮廓平稳小波变换纹理分析的彩色图像检索

摘要

由于单一特征只能表达图像的部分内容，提出了一种新的彩色图像检索方法。该算法在提取颜色特征方面，首先将图像进行分块以获得空间分布信息，为了充分利用RGB颜色模型及HSV颜色模型的优点，分别在两种不同的颜色模型中提取相应的特征向量，将两种颜色空间中的特征向量结合在一起就形成本文的颜色特征向量，在纹理特征方面，结合小波变换及轮廓波变换的优点，将图像进行非下采样轮廓平稳小波变换(NWCT)，然后分别计算各子带在各个方向上系数的均值与方差作为纹理特征向量，最后采用加权欧氏距离作为图像的相似度进行检索。实验结果表明，相对于其他检索方法，该方法平均检索精度有了一定的提高，取得了较好的检索结果。%In this paper, a new color image retrieval method based on color and texture is proposed, Firstly, in order to obtain the information of spatial distribution, images are partitioned into 5 sub-images and then the color feature is extracted from RGB model and HSV model. Secondly, each image is decomposed by stationary wavelet-nonsubsampled contourlet transform (SW-NSCT) and the mean value and standard deviation of each sub-band’s coefficient are computed as the texture features. The similarities of images are computed by the Euclidean’s distance with weight. Experiments indicate that this algorithm can improve the searching precision effectively.