摘要
颜色特征是图像检索领域中最为重要的视觉特征之一.对于基于颜色的图像检索系统,其性能在很大程度上取决于颜色特征的描述和匹配方法的有效性.本文使用主色偏差矩阵来描述颜色特征,并使用相关系数进行相似性度量.经实验证明,该方法对于图像的几何变换和空间变换,具有较高的检索查全率.
Color feature is one of the most important visual features in the field of image retrieval. For color based image retrieval system, its performance depends largely on the describing color feature and matching effecency. This paper uses dominant color difference matrix to describe the color characteristics and uses the correlation coefficient to measure similarity. The experiments show that the method has higher retrieval precision for geometrical transform and the spatial transformation of the image.
出处
《吉林化工学院学报》
CAS
2013年第11期128-131,共4页
Journal of Jilin Institute of Chemical Technology
关键词
图像检索
主色偏差矩阵
Z型扫描
相关系数
image retrieval, dominant color difference matrix, Z scan, correlation coefficient