基于差异空间相关反馈过程的高光谱图像CBIR系统  

CBIR system of hyper-spectral images based on different-spatial RF processing

作　　者：谌湘倩[1] 马绍惠[1] 王宏伟[2] 

机构地区：[1]河南机电高等专科学校计算机科学与技术系,河南新乡453002 [2]华中科技大学计算机科学与技术学院,武汉430074

出　　处：《计算机应用研究》2016年第7期2214-2218,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(61202300);河南省教育厅科学技术研究重点项目(14A520046);河南省高等学校重点科研项目(15B520006)

摘　　要：针对大多数现有的高光谱基于内容的图像检索(content-based image retrieval,CBIR)系统依赖特征表示和差异函数的问题,不完全满足传统相关反馈(relevance feedback,RF)的机器学习要求。为了解决该问题,通过差异空间的RF过程改进了CBIR系统,使用差异空间替代通常使用的特征空间,使其更适合一般机器学习技术。使用一些非零查询正样本输入到系统中,通过高光谱差异函数比较输入查询并初始化数据库获取图像的排序。最后,用户标记被系统检索的图像,选择原型图像集并开始RF过程。利用基于分离和字典的高光谱差异函数测试RF-CBIR系统,高光谱数据集的实验表明了系统的有效性,零查询的查准率和查全率接近最近RF结果。相比其他几种较为优秀的CBIR系统,系统取得了更高的查准率和查全率。As the issue of most available hyper-spectral content-based image retrieval （CBIR） systems rely on feature representation and difference function, which do not fully meet the traditional relevance feedback （RF） machine learning requirements. This paper improved CBIR system by different-spatial RF processing, and replaced feature space by differences space, making it more suitable for general machine learning techniques. Firstly, it put some non-zero positive sample querfes into the system. Then, it compared the input high spectral difference function, and initialized the database query to get the sort of the image. Finally, the user marked the retrieved system images, and selected a prototype set of images and RF process begins. It tested the RF-CBIR system by the difference function based on high-spectral separation dictionaries. Experiments on the hyperspectral data sets show the effectiveness of the system. The results of precision and recall inquiry of zero precision are closed to the best RF＇ s. The precision and the recall rate of the proposed system are higher than that of several other comparatively CBIR systems.

关 键 词：高光谱图像 图像检索 相关性反馈 差异函数 零查询 特征表示 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]