基于彩色子块共生矩阵的图像检索

Image Retrieval Based on Color Block Co-occurrence Matrix

提出了一种新型的共生矩阵描述子,它首先计算彩色图像的真彩色边缘,然后把彩色边缘图像量化为64种颜色,通过共生矩阵来描述两个彩色子块之间的空间相关性,并用于基于内容的图像检索。实验结果表明,彩色子块共生矩阵描述子能够有效地结合颜色和纹理特征,具有良好的检索性能。

This paper proposes a new kind of co-occurrence Matrix descriptor. Firstly, full-color edge is computed from color image. Secondly, the colors in image are quantized into 64 colors. Finally, describes the spatial correlation of two color blocks by co-occurrence matrix and use for image retrieval. Experiments demonstrated that the color block co-occurrence matrix descriptor considerably improves the retrieval performance and can combine the feature of color and texture effectively.