Moving Average for Error Mitigation in Pedestrian Tracking

The Moving Average is very common in the area of time series analysis mainly for analysing changes in stock market values. In this work the Moving Average will be used in another context to mitigate errors in pedestrian tracking results in video surveillance. The HOG feature extraction technique and the SVM classifier were used to detect pedestrians. For our analysis we used the database provided by the University of Edinburgh''s CAVIAR (Context Aware Vision using Image-based Active Recognition) project. The results were positive especially when the pedestrians moved vertically in the scene in the database used.