A Genetic Particle Filter for Moving Object Tracking

Abstract

Among the many techniques for tracking moving objects, the particle filter strongly attracts the researchers and developers for its simplicity and effectiveness. Two major problems suffered from the particle filter are the slow tracking speed caused by the examination of a large number of randomly generated particle samples and the position drifting of the tracked object caused by the interference from outlier particle samples. This paper proposes an improved solution to the above problems. Several experiments are conducted to verify the improvements over the conventional particle filter method.