Example-Based Color Vehicle Retrieval for Surveillance

In this paper, we evaluate several low dimensionalcolor features for object retrieval in surveillance video.Previous work in object retrieval in surveillance has beenhampered by issues in low resolution, poor segmentation,pose and lighting variations and the cost of retrieval. Toovercome these difficulties, we restrict our analysis toalarm-based vehicle detection and as a consequence, werestrict both pose and lighting variations. In addition, westudy the utility of example-based retrieval to avoid thelimitations of strict color classification. Finally, since weperform our evaluation at run-time for alarm-baseddetection, we do not need to index into a large database.We evaluate the efficiency and effectiveness of severalcolor features including standard color histograms,weighted color histograms, variable bin size colorhistograms and color correlograms. Results show colorcorrelogram to have the best performance for ourdatasets.