基于纹理和高斯密度特征的图像检索算法  

Image retrieval algorithm based on texture and Gaussian density feature

作　　者：王剑峰[1] 刘自昆[1] 

机构地区：[1]重庆电子职业技术学院计算机系,重庆400021

出　　处：《计算机工程与设计》2008年第19期4995-4998,共4页Computer Engineering and Design

摘　　要：直接从DCT域中提取图像的特征是提高图像的检索效率的方法。直接从压缩域中提取图像的高斯密度,即计算图像在8个方向上的分段累加值,形成一个8*4的二维向量,再结合图像的纹理特征来进行图像检索。为了验证算法的可行性,建立了10000幅图像的图像库。实验结果表明,该方法能够准确地检索出目标图像,有效地提高了图像检索的精度和速度。To improve efficiency of the image retrieval, the techniques of the direct feature extraction in DCT domain are extensively emphasized. A new algorithm for compressed image retrieval is proposed based on Gaussian density feature （GDF）. This algorithm directly extract Gaussian density of 8 direction from compressed image data to construct a 2-dimention array （8＊4） as an indexing key to retrieve images based on their content features and texture feature. To test and evaluate the proposed algorithms, experiments with a database of 10 000 images are carried out. In comparison with existing representative techniques, the experimental results show the su- periority of the proposed method in terms of retrieval precision and processing spbed.

关 键 词：基于内容的图像检索 高斯密度特征 灰度共生矩阵 离散余旋变换 查准率 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]