您的位置:首页 > 论文页面
基于多特征融合的乳腺图像哈希检索方法
发表时间:2016-11-30 浏览量:1957 下载量:562
全部作者: | 吕鑫,王颖,刘璐 |
作者单位: | 西安电子科技大学电子工程学院 |
摘 要: | 为提高计算机辅助检测(computer aided detection,CAD)系统性能,帮助医生进行更精确的诊断,提出一种基于多特征融合的乳腺图像哈希检索方法。该方法通过对乳腺图像中可疑区域提取多种不同的特征:多维度词袋模型(pairwise bag of words,PBoW)特征、分层加权Gist 特征及方向梯度直方图(histogram of oriented gradient,HOG)特征等,更全面地表征肿块图像内容信息。将融合特征与哈希算法结合,并引入图模型理论,实现对可疑区域的概率预测,得到基于多特征融合哈希的检索结果。实验结果表明,该方法能够准确预测可疑区域是否为肿块的概率,降低假阳性率的同时,实现检测系统整体性能的提升。 |
关 键 词: | 图像处理;图像检索;多特征融合;图模型;概率预测 |
Title: | Mammography hashing retrieve based on multi-feature fusion |
Author: | LÜ Xin, WANG Ying, LIU Lu |
Organization: | School of Electronic Engineering, Xidian University |
Abstract: | To improve the performance of computer aided detection (CAD) systems and assist doctors to make more accurate diagnosis, in this paper, we proposed a multi-feature fusion retrieval algorithm based on hashing. To fully describe breast image, different specific features of every suspicious region are extracted, such as pairwise bag of words (PBoW) and hierarchy-weigh Gist and histogram of oriented gradient (HOG) features. We integrate multi-feature fusion and hash algorithms to predict the probability of suspicious regions by the graph model. The experimental results show that the method proposed in this paper can accurately predict probability of suspicious regions, and improve performance of the detection system while maintaining a low false positive rate at the same time. |
Key words: | image processing; image retrieval; multi-feature fusion; graph model; probability forecast |
发表期数: | 2016年11月第22期 |
引用格式: | 吕鑫,王颖,刘璐. 基于多特征融合的乳腺图像哈希检索方法[J]. 中国科技论文在线精品论文,2016,9(22):2337-2347. |

请您登录
暂无评论