Haberlerde kişileri bulmaya yarayan çizgeye dayalı bir yöntem

Along with the recent advances in technology, large quantities of multi-modaldata has arisen and became prevalent. Hence, effective and efficient retrieval,organization and analysis of such data constitutes a big challenge. Both newsphotographs on the web and news videos on television form this kind of data bycovering rich sources of information. People are mostly the main subject of thenews; therefore, queries related to a specific person are often desired.In this study, we propose a graph based method to improve the performanceof person queries in large news video and photograph collections. We exploitthe multi-modal structure of the data by associating text and face information.On the assumption that a person?s face is likely to appear when his/her name ismentioned in the news, only the faces associated with the query name are selectedfirst to limit the search space for a query name. Then, we construct a similaritygraph of the faces in this limited search space, where nodes correspond to thefaces and edges correspond to the similarity between the faces. Among thesefaces, there could be many faces corresponding to the queried person in differentconditions, poses and times. There could also be other faces corresponding toother people in the news or some non-face images due to the errors in the facedetection method used. However, in most cases, the number of correspondingfaces of the queried person will be large, and these faces will be more similar toeach other than to others. To this end, the problem is transformed into a graphproblem, in which we seek to find the densest component of the graph. Thismost similar subset (densest component) is likely to correspond to the faces ofthe query name. Finally, the result of the graph algorithm is used as a model forfurther recognition when new faces are encountered. In the paper, it has beeniiishown that the graph approach can also be used for detecting the faces of theanchorpersons without any supervision.The experiments are performed on two different data sets: news photographsand news videos. The first set consists of thousands of news photographs fromYahoo! news web site. The second set includes 229 broadcast news videos pro-vided by NIST for TRECVID 2004. Images from the both sets are taken in reallife conditions and, therefore, have a large variety of poses, illuminations andexpressions. The results show that proposed method outperforms the text onlybased methods and provides cues for recognition of faces on the large scale.Keywords: Face recognition, face retrieval, SIFT features.