Sub-difference Image of Curvelet Transform for Crowd Estimation: A Case Study at the Holy Haram in Madinah

Counting people and estimating their densities over a certain area is a fundamental task for many artificial intelligence systems. In this study, sub-difference images of curvelet transform are postulated as an efficient source for effective crowd estimation features. The new algorithm is described in detail in the form of a case study conducted at the Holy Haram in Madinah. The application of the difference images extracted by curvelet transform is thus proven to be efficient and useful for further studies. In addition, the proposed method is independent of any background modeling or background subtraction techniques. The method can also handle crowds of different sizes and strong perspective distortion conditions. The estimation procedure is performed using two versions of difference images generated by forward and customized inverse curvelet transforms. The proposed algorithm is then compared with normal difference image utilization.