Alpha Matting Using Artificial Immune Network

作者：Hao, Zhifeng^*; Liu, Jianming; Yan, Xueming; Wen, Wen; Cai, Ruichu

来源：3rd International Conference on Swarm Intelligence (ICSI), Shenzhen, PEOPLES R CHINA, 2012-06-15 To 2012-06-20.

DOI：10.1007/978-3-642-30976-2_36

• alpha matting
• Artificial Immune Network
• feature map

Alpha matting refers to the problem of softly extracting the foreground from an image. To solve the matting problem initialized with a trimap (a partition of the image into three regions: foreground, background and unknown pixels), an approach based on artificial immune network is proposed in this paper. The method firstly uses Artificial Immune Network(aiNet) to map the color feature for unknown region, attaining the color subset both on the foreground and background color distributions, then estimate the alpha matte for unknown region, and finally apply guided filter to improve the matting results. Experiments on several different image data sets show that the proposed method produces high- quality matting results.

• 出版日期2012
• 单位计算机软件新技术国家重点实验室; 广东工业大学