Research on Sampling Methods in Particle Filtering Based upon Microstructure of State Variable

  • Authors:

  • Zhiquan Feng;Bo Yang;Yuehui Chen;Yanwei Zheng;Yi Li;Zhonghua Wang

  • Affiliations:

  • School of Information Science and Engineering, University of Jinan, Jinan, P.R. China 250022;School of Information Science and Engineering, University of Jinan, Jinan, P.R. China 250022;School of Information Science and Engineering, University of Jinan, Jinan, P.R. China 250022;School of Information Science and Engineering, University of Jinan, Jinan, P.R. China 250022;School of Control Science and Engineering, University of Jinan, Jinan, P.R. China 250022;School of Control Science and Engineering, University of Jinan, Jinan, P.R. China 250022

  • Venue:
  • ICIC '08 Proceedings of the 4th international conference on Intelligent Computing: Advanced Intelligent Computing Theories and Applications - with Aspects of Theoretical and Methodological Issues
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

With the purpose of decreasing the number of particles needed to be sampled stochastically in particle filtering, a new sampling method used in particle filtering is put forward in this paper. First of all, under specific human-computer conditions, both cognitive psychology features of operators and motion features of the operators' hands are studied, upon which a novel concept, microstructure of state variable (MSV), is proposed. Then, a general method to describe the MSV is further discussed. At last, a Sampling Algorithm based on the MSV is put forward, which is also called SAMSV. The MSV provides an unique and efficient mathematic model that makes SAMSV avoid sampling huge particles with poor quality. In order to demonstrate validity and performance of SAMSV, a great deal of experiments is completed. By using fewer particles, compared with conventional particle filtering, the SAMSV may achieve better tracking precision. In forms of both the theoretical analysis and a great deal of experimental results from real video data, just a few particles are needed to describe post probability distribution of state variables without reducing the tracking precision by our dimensionality reduction method.