摘要
对海量图片数据信息进行检索的研究,对于增加多媒体图片信息检索率具有重要意义。由于海量多媒体图片信息存在一定的相似特征,使得进行检索时图片信息特征出现混乱的情况。传统的图像检索方法,主要通过图片信息特征进行分类检索,当图片信息混乱时,无法对海量多媒体图片信息特征进行分类,导致检索速度慢、准确率低的问题。提出一种新的海量多媒体图片信息高效检索算法,通过分析图片的SIFT局部特征,获取图片SIFT特征,再利用SIFT特征进行图片匹配对目标图像进行识别,引入BOW算法对匹配后的图片SIFT特征进行索引,建立Bag of words模型、TF-IDF加权和欧式距离来完成图片的相似度计算,完成海量多媒体图片信息的检索。仿真结果表明,采用改进的检索方法有效的提高了检索速度和检索准确率。
Research on the retrieval algorithm for massive image data is of great significance to increase the rate of multimedia image information retrieval. In the paper, a new efficient retrieval algorithm for massive multimedia image information was proposed. Through the analysis of the SIFT local features of the image, the image SIFT features were obtained, and then the SIFT features were used for image matching to identify the target image. BOW algorithm was introduced to index the SIFT features of the matching image and build a Bag of words model, TF-IDF weighting and Euclidean distance to complete the similarity calculation of the image, thereby the massive muhimedia image information retrieval was completed finally. The simulation results show that the improved retrieval method can effectively improve the retrieval speed and accuracy.
出处
《计算机仿真》
CSCD
北大核心
2016年第11期280-283,389,共5页
Computer Simulation
关键词
多媒体图像
信息
检索
Multimedia image
Information
Retrieval
