Abstract: This paper proposes an image retrieval method based on binary color co-occurrence matrix named Binary Modified Color Co-occurrence Matrix(BMCCM). It uses “color invariant model of RGB quantity” to quantify the original image, calculates the color co-occurrence matrix of transformed image, gets the binarization of co-occurrence matrix, and uses Jaccard coefficient to measure the similarity between images to complete the image retrieval. Experimental results show that BMCCM is more accurate and efficient on retrieval than color histogram and color (auto) correlogram.

Key words: color co-occurrence matrix, color invariant, image retrieval, color correlogram

摘要： 提出一种基于二进制颜色相关矩阵的图像检索方法BMCCM。该方法利用RGB分量颜色不变量模型对原始图像进行量化，计算变换后图像的颜色相关矩阵，对相关矩阵进行二值化，并利用Jaccard系数衡量图像特征间的相似度，完成图像的检索。实验结果表明，BMCCM的检索准确度及效率明显高于颜色直方图和颜色(自)相关图。

关键词: 颜色相关矩阵, 颜色不变量, 图像检索, 颜色相关图

CLC Number: 

• TP311.13