Background Modeling and Subtraction Using a Local-linear-dependence-based Cauchy Statistical Model

Many motion object detection algorithms rely on the process of background subtraction, an important technique which is used for detecting changes from a model of the background scene. The background model affects object detecting algorithm tolerating changes in background scene and the granularity of the detected foreground objects. An algorithm using a subtracted background modeling based on Cauchy statistical distribution the purpose of object detecting is presented. The paper concludes that the ratios of the intensity values between background image and current image are fitted to a Cauchy distribution. The Cauchy has much heavier tails and better represents the tails of the histogram than the Gaussian. The Cauchy based method without exponential operation is more cost-efficient than the Gaussian. The proposed approach takes the advantages of the statistic distribution characteristic of pixels and spatial correlativity of the region around a pixel to subtract background. The paper also discusses the changes in background scene in detail. At last, a robust object detecting approach being invariant or adapting to the changes in background scene is acquired by hypothesis test. Experimental results demonstrate the proposed algorithms can tolerate the whole or local sudden or slow change in illumination, filter clutter noises caused by small motion in background scene, and adapt to rain