Perceptual Color Space Based Natural Image Matting

This paper proposes a fast natural image matting method based on perceptual color space. This method begins from segmenting regions by a user, and then uses a practical model to estimate the foreground and background color components of each unknown point. Last, we present a novel perception based alpha estimating approach by estimating chroma and intensity alpha respectively. This method is much faster than the existing approaches of comparable quality.