Background Modeling
Maheshkumar H. Kolekar · 2018
This chapter explains the statistical background modeling methods based on Gaussian mixture model (GMM). Background modeling is a technique used in computer vision for the detection of foreground objects in a frame sequence. The accuracy of foreground object detection depends on the accuracy in the modeling of the background. Background modeling techniques can be broadly classified based on non-statistical and statistical approaches. In the non-statistical approaches, the first frame is considered to be the background and the subsequent frames are subtracted from the background. In the statistical approaches, the probability distribution functions of the background pixels are estimated. Improvements in background modeling have led to numerous applications, such as event detection, object behavior analysis, suspicious object detection, and traffic monitoring. The chapter discusses the shadow detection and removal techniques. The performance of many video surveillance algorithms, such as object segmentation, object detection, object classification, and object tracking degrade because of shadows of various objects in the scene.