Image matting is a fundamental technique used in many image and video applications. It aims to softly extract foreground from the image accurately. In this paper, we propose a new matting approach based on naive Bayes classifier to produce matting results with higher accuracy. Spatially-varying probabilistic models for the classifier are established. Confidence values are defined to make better use of the classification results. The results are then refined and combined with closed-form matting to obtain the final alpha matte. We conduct qualitative and quantitative evaluations. Results show that our method outperforms many recent algorithms. © 2012 Springer-Verlag.
CITATION STYLE
Zhang, Z., Zhu, Q., & Xie, Y. (2012). A novel image matting approach based on naive Bayes classifier. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7389 LNCS, pp. 433–441). https://doi.org/10.1007/978-3-642-31588-6_56
Mendeley helps you to discover research relevant for your work.