Generalized motion and edge adaptive interpolation de-interlacing algorithm

Abstract

This paper presents a generalized motion and edge adaptive de-interlacing framework, which offers a structured way to develop de-interlacing algorithm. The framework encompasses many typical de-interlacing algorithms, ranging from simple interpolation based algorithms, to more complex edge dependent and motion adaptive algorithms. Based on this framework, we develop a new de-interlacing algorithm which is efficient and artifacts-free. The proposed algorithm was evaluated by five video sequences, namely, "Akiyo", Mother and Daughter", "Silent", "Foreman" and "Stefan". Experimental results confirm that the proposed algorithm performs, both objectively and subjectively, much better than other similar algorithms. These promising results indicate that the proposed framework has good potential for realizing even better de-interlacing algorithms.