基于统计特征的DCT压缩域纹理图像检索方法  

Algorithm of texture images retrieval in DCT compressed domain based on statistical features

作　　者：张曦煌[1] 卞国春[1] 李红[1] 

机构地区：[1]江南大学信息工程学院,江苏无锡214122

出　　处：《计算机工程与设计》2006年第7期1282-1285,共4页Computer Engineering and Design

摘　　要：提出了一种基于离散余弦变换(Discrete Cosine Transfrom,DCT)的纹理图像的检索方法。该方法在DCT压缩域,通过直接对DCT系数计算,获得图像纹理的统计特征,并作为检索的依据。理论分析和实验结果都表明,该方法具有很好的检索准确率和效率,并且对于旋转具有不变性。A new algorithm of retrieval for texture images in DCT compressed domain is presented. The statistical features of the texture images is computed directly from DCT coefficients, and used for image retrieval. Theoretical analysis and experimental results show that the approach achieves good performances in terms of retrieval accuracy and effectiveness. It is also invariant to rotation.

关 键 词：DCT变换 压缩域 纹理 统计特征 图像检索 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]