【摘要】 在彩色图像检索中 ,颜色是最重要的特征 .提出了一种基于HSV颜色模型的图像检索算法 ,该算法根据人类视觉系统对颜色感知的经验值对图像的色调、亮度和饱和度进行非均匀量化 ,并将量化后得到的 2 5 6柄一维矢量展开成 16× 16的特征矩阵 ,通过线性变换的方法求出特征矩阵的特征值 ,构成仅有16个元素的特征矢量 ,从而有效地压缩了特征矢量的维数 ,减小了特征数据库的规模 ,提高了图像的在线查询速度 .同时 ,为提高图像查准率 ,采用了基于全局和局部颜色特征的分层查找策略 .试验表明 ,根据用户给出的示例图像 ,通过两层查询技术能够获得较好的检索效果更多还原

【Abstract】 Content based image retrieval is a hot research subject. In color image retrieval, color is the most important feature. An algorithm for image retrieval based on HSV color model is proposed in this paper. The algorithm non uniformly quantifies the hue, value and saturation of images with the experiential data of color sensation produced by the human visual system, expands the quantified 256 bin one dimensional vector into a 16× 16 eigenmatrix, evaluates the eigenvalues of the eigenmatrix by means of linear transformation, and finally forms a 16 element feature vector. So the number of dimensions of the feature is effectively reduced, the size of the feature database is decreased, and the speed of on line query is increased. Meanwhile, in order to perfect the query precision, the hierarchical query method indexed by global and local color features is adopted. As shown in the experiments, with an example image given by the user, good retrieval results can be obtained by two level query technique with color features.更多还原