摘要
提出一种在压缩域中基于随机变量数字特征累加直方图的图像检索算法,利用离散余弦变换(discrete cosine transform,DCT),计算图像块系数的数学期望和方差,通过期望和方差映射到坐标系的概率建立累加直方图并产生图像的特征向量,利用相关系数判定查询图像和数据库图像的差异。实验结果显示,此种方法检索精度高、速度快、计算量较小。
This paper puts forward the image retrieval algorithm based on the numerical characteristics of random variables histogram in the compressed domain, the use of discrete cosine transform( discrete cosine transform, DCT) , and the mathe- matical calculation of the image block expectation and variance, through the expectation and variance of the probability map to coordinate the establishment of cumulative histogram Figure and produce images of the feature vectors and regions. The similarity between query image and database image is computed using the correlation coefficient. The experiment is carried oil two data sets. The result shows this method can achieve more effective retrieval precision.
出处
《重庆邮电大学学报(自然科学版)》
CSCD
北大核心
2013年第5期700-704,710,共6页
Journal of Chongqing University of Posts and Telecommunications(Natural Science Edition)
基金
2010年重庆市高等教育教学改革基金(101404)
2010年中国航天科技集团第七研究院青年科技创新基金([2010]776)
2011年重庆市科学技术研究基金(KJ113201)~~
关键词
离散余弦变换
图像检索
直方图
数字特征
压缩域
discrete cosine transform
image retrieval
histogTam
numerical characteristics
compressed domain