CAJ | 학술논문

图像集匹配是模式识别领域研究的热点问题之一。从图像分布的局部结构出发，提出格拉斯曼流形上局部结构保持的图像集匹配方法。将图像集合张成的子空间投影到格拉斯曼流形，通过子空间之间的典型相关计算格拉斯曼核，将集合的相似性转换为流形上点之间的距离的计算。在基于图像集合的对象识别任务上测试提出的算法，实验结果表明，提出的方法在识别率上超越了当前主流的图像集匹配算法。
도상집필배시모식식별영역연구적열점문제지일。종도상분포적국부결구출발，제출격랍사만류형상국부결구보지적도상집필배방법。장도상집합장성적자공간투영도격랍사만류형，통과자공간지간적전형상관계산격랍사만핵，장집합적상사성전환위류형상점지간적거리적계산。재기우도상집합적대상식별임무상측시제출적산법，실험결과표명，제출적방법재식별솔상초월료당전주류적도상집필배산법。
Image set matching is one of the research focuses in the field of pattern recognition.In this paper we propose an image set matching method with locality structure preserving on Grassmann manifold proceeding from the locality structure of image distribution.The subspaces spanned by the image set are mapped onto the Grassmann manifold,and the canonical correlation between subspaces is used to calculate the Grassmann kernel,then the similarity between the image sets is converted into the calculation of the distance between two points on manifold.The proposed method is tested in image set-based object recognition tasks.Experimental results show that the proposed method outperforms current mainstream image set matching methods in terms of recognition rate.