Pedestrian Identification, Tracking, and Counting with Shadow Detection in Monochromatic Video Streams

C. Bräuer-Burchardt (Germany)

Keywords

Tracking, Shadow Detection, Identification, Image Sequences

Abstract

A new robust method to count walking persons in busy public areas in using vertically fixed monochromatic cam eras is presented. Shadows and not separated persons are the main reasons for errors in pedestrian counting. That's why the focus is directed to shadow detection and object separation. The main novelty of the method is a new approach to shadow detection. The information about existence and kind of shadows from the former images of the sequence is used to predict the properties of the shadows in the current image. Two different methods to detect weak and strong shadows are presented.

Important Links:



Go Back