CAJ | 학술논문

在图像检索领域应用商空间粒度思想,阐述了图像检索中的粒度原理,对图像检索问题进行商空间描述,并构造等价关系,提出了一种新的图像检索方法.根据图像在不同粒度下的表现构造不同粒度下的商空间,然后根据商空间粒度合成原理构造最优准则函数,对已得到的商空间进行属性合成,进而以合成后的属性函数来完成图像检索.与采用单一属性特征的图像检索方法相比,该方法能够在更大程度上利用图像自身所提供的信息,取得更好的检索效果和更高的查全率查准率指标.
재도상검색영역응용상공간립도사상,천술료도상검색중적립도원리,대도상검색문제진행상공간묘술,병구조등개관계,제출료일충신적도상검색방법.근거도상재불동립도하적표현구조불동립도하적상공간,연후근거상공간립도합성원리구조최우준칙함수,대이득도적상공간진행속성합성,진이이합성후적속성함수래완성도상검색.여채용단일속성특정적도상검색방법상비,해방법능구재경대정도상이용도상자신소제공적신식,취득경호적검색효과화경고적사전솔사준솔지표.
Applying quotient space granularity thinking in image retrieval field,granularity principle in image retrieval is clarified,quotient space description is done for image retrieval,equivalent relation is constituted,and a new image retrieval method is proposed.Different quotient spaces are constituted according to different behaviors under different granularities of image,and optimal criterion function is also constituted to do the attribute combination for the achieved quotient spaces.With the attribute combination,image retrieval is accomplished.Compared with methods adopting single attribute feature,this method can make use of the image information with its own in a more effective way and achieve better retrieval effect and higher recall precision guideline.