关键词: 实时运动检测, 背景模型, 高斯混合模型, 均匀混合模型, 学习率自适应调整

Abstract: There are numerous approaches to Real-time segmentation of moving regions in image sequences. A typical method is background subtraction. One of the successful solutions to these problems is to use a gaussian mixture model (GMM) per pixel. However, the method suffers from fixed learning rate. In addition, it requires excessive computational resources or time in busy environments. This paper presents a method which adopts uniformity mixture model with adaptive learning rate. This allows our system learn faster and more accurately as well as adapt effectively to changing environments. A comparison has been made. The results show the speed of learning and the accuracy of the model using our method over the gaussian mixture model.

Key words: Real-time motion detection, background model, gaussian mixture model, uniformity mixture model, adaptive learning rate