公开(公告)号：KR1020010054460A

公开(公告)日：2001-07-02

申请号：KR1019990055279

申请日：1999-12-06

Applicant: 한국전자통신연구원

Inventor： 김영태 ,  이호영 ,  민운기 ,  김진웅 ,  하영호 ,  노용만

IPC: G06T5/40

CPC classification number: G06F17/30781 ,  G06T5/40 ,  G06T7/90

Abstract: PURPOSE: An image indexing and retrieving method using color information is provided, which realizes the image retrieval of high reliability by removing the color distribution variance, recovering the inherent color of an object, and extracting/indexing the spatial position and distribution information of similar colors. CONSTITUTION: First, the color variance width of a data image is adjusted so that the variance of color distribution generated due to the variance of a beam irradiation direction can be removed if the data image to be indexed is input to a database. Then, color histogram is calculated and indexed by quantizing the data image of which the color variance width is adjusted. Next, the color variance width of an input inquiry image is adjusted. Then, color histogram is calculated by quantizing the inquiry image of which the color variance width is adjusted. Finally, the similarity between the color histograms of the inquiry image and the data image is measured to be retrieved.

Abstract translation: 目的：提供使用颜色信息的图像索引和检索方法，通过去除颜色分布方差，恢复对象的固有颜色，以及提取/索引相似颜色的空间位置和分布信息，实现高可靠性的图像检索 。 构成：首先，调整数据图像的颜色方差宽度，使得如果将要索引的数据图像输入到数据库，则可以消除由于光束照射方向的变化而产生的颜色分布的方差。 然后，通过量化调整颜色方差宽度的数据图像来计算和索引颜色直方图。 接下来，调整输入查询图像的颜色方差宽度。 然后，通过量化调整颜色方差宽度的查询图像来计算颜色直方图。 最后，测量查询图像的颜色直方图与数据图像之间的相似度。

公开(公告)号：KR100356018B1

公开(公告)日：2002-10-18

申请号：KR1019990055279

申请日：1999-12-06

Applicant: 한국전자통신연구원

Inventor： 김영태 ,  이호영 ,  민운기 ,  김진웅 ,  하영호 ,  노용만

IPC: G06T5/40

Abstract: 본발명은정지영상으로구성된데이터베이스를대상으로원하는영상을신속, 정확하게검색하기위하여영상의광원조사방향의변화로인한밝기값변화를제거하고, 유색광원의영향으로인한물체색의변화를제거하여물체고유색을복원하며, 다중양자화를통하여정확한색 히스토그램특징벡터를추출하고, 유사색의공간적인위치및 분포정보를추출및 색인화하는색 정보를이용한영상색인방법및 검색방법에관한것이다. 이를해결하기위해본 발명의색 정보를이용한영상색인방법은, 광원조사방향의변화로인해발생하는밝기(intensity) 변화즉, 색분포(color distribution) 변화를제거하기위해영상의생동폭을조절하는제 1 단계와; 유색광원의영향으로인한물체색의변화를제거하여고유의물체색을복원하기위하여색 보정을하는제 2 단계; 다중양자화를통한색 히스토그램을산출하는제 3 단계; 양자화를통해얻어진색 히스토그램의동일빈에속한화소들의공간적인정보를구하는제 4 단계; 및추출된특징벡터를색인화하는제 5 단계를포함한다.