Abstract—Estimating crowd density may be a good solution for control management, maintaining the crowd safety, or prevention of riot and high risk activities. This paper presents a computational fast and simple method for estimating crowd density based on histogram model classification. The histogram model here is based on the proposed Improved Uniform Local Binary Pattern features. Two main advantages of using this improved version of the local binary pattern are that the pattern features are now intensity invariant as well as rotational invariant. Our proposed method also uses less number of features which makes it faster without sacrificing the overall performance. It has been shown that this method is robust in areas with very low, low, and medium crowd densities. Performance and comparisons with the original local binary pattern method are demonstrated in experimental results.

Index Terms—Crowd estimation, histogram model classification, binary pattern

The authors are with Faculty of Electrical Engineering, Universiti Teknologi Malaysia, 81310, Skudai, Johor, Malaysia (email:mosavi@fkegraduate.utm.my,shahdi@fkegraduate.utm.my,syed@fke.utm.my)

Cite: Seyed Mojtaba Mousavi, Seyed Omid Shahdi, and S. A. R. Abu-Bakar, "Crowd Estimation Using Histogram Model Classification Based on Improved Uniform Local Binary Pattern," International Journal of Computer and Electrical Engineering vol. 4, no. 3, pp. 256-259, 2012.