Title :
Towards optimal ranking metrics
Author :
Sebe, N. ; Lew, M. ; Huijsmans, D.P.
Author_Institution :
Dept. of Comput. Sci., Leiden Univ., Netherlands
Abstract :
Euclidean metric is frequently used in computer vision, mostly ad-hoc without any justification. However we have found that other metrics like double exponential metric or Cauchy one provide better results, in accordance with the maximum likelihood approach. In this paper we experiment with different modeling functions for similarity noise and compute the accuracy of different methods using these modeling functions in three kinds of applications: content-based image retrieval from a large database, stereo matching and video sequences. We provide a way to determine the modeling distribution which fits best the similarity noise distribution according to the ground truth. In the optimum case, when one has chosen the best modeling distribution, its corresponding metric will give the best ranking results for the ground truth provided
Keywords :
computational geometry; computer vision; image matching; image sequences; information retrieval; maximum likelihood estimation; visual databases; Euclidean metric; computer vision; content-based image retrieval; double exponential metric; large database; maximum likelihood approach; optimal ranking metrics; ranking results; similarity noise distribution; stereo matching; video sequences; Content based retrieval; Digital images; Feature extraction; Histograms; Image databases; Image retrieval; Information retrieval; Spatial databases; Stereo vision; Video sequences;
Conference_Titel :
Computer Graphics, Image Processing, and Vision, 1998. Proceedings. SIBGRAPI '98. International Symposium on
Conference_Location :
Rio de Janeiro
Print_ISBN :
0-8186-9215-4
DOI :
10.1109/SIBGRA.1998.722776
