Introduction to Image Matting

Matting refers to the problem of accurate extraction of the foreground object in images and video, which plays a fundamental role in many image and video editing operations. In this talk, an introduction to the image matting problem is presented. The talk covers the basic alpha matting problem and some natural image matting algorithms, such as closed form matting, robust and improved color sampling methods. Furthermore, a web-based performance evaluation system (www.alphamatting.com) developed by members of our group is presented. In this context, the acquisition of ground truth datasets and perceptually motivated error metrics are discussed in more detail.