中文摘要：

针对背景建模方法对场景光照变化较为敏感，提出融合时间预测和空域纹理信息设计全新的时空一对称ST—CSLBP算子，构建基于ST—CSLBP直方图的混合时空背景模型，融合时域分布统计法和空域背景各自的优势，以同时适应长时光照以及短时的光照变化。实验表明，该算法与现有背景建模算法相比，具有较好的准确率和检索率。

英文摘要：

Considering that background modeling algorithms are quite sensitive to luminance change in scene, in this paper, a novel space-time symmetrical ST-CSLBP operator is proposed, which integrates time prediction and texture information. And a new mixed space-time background model based on ST-CSLBP histogram is built, which merges the advantages of time-domain statistics and spatial distribution, and adapts to long and short luminance changes. Compared with existing background modeling algorithm, this method has better accuracy and retrieval rate.