一种用于运动物体检测的自适应更新背景模型

Adaptive Update Background Model for Detecting Moving Objects

提出一种能自适应更新的背景模型,使得运动物体检测中场景变化时提取的背景能进行相应的更新.该模型利用特征点信息将场景中的像素点分类,并针对不同类别像素点计算其更新速率;然后对像素点采用相应速率进行更新,从而能根据场景不同变化进行自适应更新.实验结果表明,该模型能较好地处理混合高斯模型因采用同一更新速率导致的背景模型更新错误问题.

An adaptive background updating model is proposed to tackle possible background changes in moving object detection. The model uses feature points to classify the pixels in the scene, then compute their updating rates according to their classification information. Finally, background is updated adaptively with varying update rate. The experimental results show that the proposed model can effectively cope with the inaccurate updating problem in the Gaussian mixture model due to the fixed update rate.