双背景模型的快速鲁棒前景检测算法

针对前景检测中的光照变化问题,提出一种基于双背景模型的快速鲁棒前景检测算法.通过建立简单的快慢双背景模型,提高前景检测的效率.结合视频时间感知信息和光照补偿措施,增强算法对光照变化的鲁棒性,提高前景检测精度.在具有光照变化的公开数据集上进行测试,实验结果表明:所提出的算法不仅对光照变化有较强的鲁棒性,同时,具有极快的处理速度.

Fast and Robust Foreground Detection Algorithm Based on Double Background Model

In order to solve the problem of illumination changes in foreground detection, a fast and robust foreground detection algorithm based on double background model was proposed in this paper. The efficiency can be improved by establishing a simple double background model with fast and slow update rate; the robustness with illumination variations can be enhanced by combining with video time perception information and illumination compensation measures, and hence improving the precision of the foreground detection. Experiments were performed on several challenging sequences with illumination variations in the benchmark evaluation, and the results show that the proposed algorithm not only owns good robustness with changing of illumination, but also has very fast process speed.