Counting people by extracting low level features using Bayesian Regression

Abstract: Counting people is a crucial component in visual surveillance mainly for crowd monitoring and management. Now days, significant improvement has been made on the field by using features regression. On this context, perspective distortions have been frequently studied; however, crowded scenes remain particularly challenging and could extremely affect the count because of the partial occlusions that take place between individuals. To overcome this challenge this paper proposes Gaussian Mixture Model (GMM). It uses background subtraction to obtain highly accurate foreground segmentation for people counting approach. It leads to integrating an uniform motion model. The counting is established on foreground measurements, where perspective normalization and a crowd counting measures the density of a crowd with foreground pixel counts into a single feature. Afterwards, the correspondence between this frame-wise feature such as internal features, segment features, texture features are extracted and these extracted features are applied to the Bayesian regression. Then count is estimated in each frame.

Keywords: Bayesian regression, crowd analysis, Gaussian Processes