A robust foreground detection method for local lighting changes via adaptive LBP and Gaussian mixture fusion
Weiwei Song, Yongda Zhang, Chenyang Zhu, Zhengkang Yan, Lijun Yin · 2025
This article focuses on the foreground detection under local light mutation. The main work is as follows: In response to the sensitivity of existing foreground detection algorithms to sudden changes in lighting in the scene, a fusion improved LBP algorithm and mixture Gaussian model algorithm are proposed. This algorithm model integrates spatial and temporal information, adds central pixel information, and introduces an adaptive lighting factor based on the LBP algorithm, so that the threshold can adaptively change with different lighting conditions. At the same time, the Gaussian model algorithm and LBP algorithm complement each other's shortcomings in foreground detection, allowing our algorithm to solve the "hole" problem in Gaussian model detection while retaining the accuracy of LBP algorithm in detecting edge information. This article validates the effectiveness of the algorithm through experiments on commonly used experimental scenarios and special scenarios with local lighting mutations.