Automatic Image / Video Cutout & Matting

Most existing techniques of extracting foreground work only in interactive mode. This paper introduces a novel algorithm of automatic foreground extraction for special object, and verifies its effectiveness with head shoulder pictures. The main contribution of our idea is to make the most of the prior knowledge to constrain strongly the processing of foreground extraction. For human head shoulder images, we first detect face and a few facial features, which help estimate an approximate mask covering the interesting region, based on the Candide face model (which is the subset of MPEG4 face standard model); the algorithm then extracts the hard edge of foreground from the specified area, using an iterative graph cut method incorporated with an improved Gaussian Mixture Model. To generate accurate soft edges, a Bayes matting is employed. The whole process is fully automatic. Experimental results demonstrate that our algorithm is both robust and efficient.

The upper row shows four original head shoulder pictures and the lower row presents the corresponding cutout results.

From left to right. The original picture; the cutout result;  and the composite result.

Matting ( Creating a "soft" edge around the cutout "hard" edge).