Authors: Sweta Singh; Anand Singh Jalal
Addresses: GLA University, Mathura-281406, India ' GLA University, Mathura-281406, India
Abstract: Image matting is a process of extracting a foreground object from an image. Providing trimap is an important challenge of image matting. An attentively created trimap is required to get accurate alpha-matte. Accuracy of trimap-based approach reduces drastically as the trimap becomes thicker and coarser. Providing trimap manually is an exhausting and time consuming task for user. Scribble-based approach solves the problem by reducing user effort but with reduced accuracy. In this paper, we propose an approach to automatically generate trimap. It uses learning-based matting to generate a high quality alpha-matte in less time. It helps in obtaining accurate and fast result without user effort. The experimental results show the automatically generated trimap and alpha-matte are qualitatively and quantitatively better.
Keywords: alpha-matte; image matting; opacity; partial pixel; scribble; trimap generation; object extraction; foreground objects; learning-based matting; digital images; image processing.
International Journal of Machine Intelligence and Sensory Signal Processing, 2014 Vol.1 No.3, pp.232 - 250
Received: 20 Jun 2013
Accepted: 10 Oct 2013
Published online: 19 Dec 2014 *