Digital Image Cutout Algorithm Research

Digital image matting is an image processing technology which can preciselyextract any specified foreground object in natural scene image. For example, it enablesdirectors to transport an actor into a completely new location. Nowadays matting iswidely used in magazines,2D arts, advertisement and video post production. It is acrucial and frequently used operation in visual effects production.Matting has been developed several decades ago, but it hasn’t been extensivelystudied until recent years the computer and multimedia technique rapidly developing.Traditional matting methods are designed to solve a restricted version of the generalproblem in studio. Natural image matting can be used to process images with anynatural background, but such methods have high computational complexity and can notguarantee the accuracy.This article is for the aforementioned problems of matting technology, carried outthe following work:1. The current digital matting approaches, such as natural image matting,environment matting, shadow matting and video matting, are systematically studied.Three typical matting approaches, such as Robust matting, Bayesian matting andClosed-Form matting are experimented.2. Examplar-Based Inpainting technique is studied and added to the mattingalgorithm. This background recovery prior based matting method can improve the maskaccuracy without raising the computational complexity. This method can get theforeground object removed digital image as the by product of getting the matting mask.3. The relationship between data sampling and image gradient is studied and animage patch gradient direction prior based sampling strategy is designed. Combined thisnew sampling strategy with the color line model, a new matting method is designed.Experimental results show that this method has good matting mask accuracy.