Motion-compensated TV-reconstruction-based video deinterlacing. (1st January 2014)

• Record Type: Journal Article
• Authors: Trocan, Maria
• Is Part Of: International journal of intelligent information and database systems. Volume 8:Number 3(2014)
• Page Start: 209
• Page End: 223
• Publication Date: 2014-01-01
• Subjects: deinterlacing -- motion compensated interpolation -- sparse reconstruction; Database management -- Computer programs -- Periodicals; Information retrieval -- Computer programs -- Periodicals; Information storage and retrieval systems -- Computer programs -- Periodicals; Artificial intelligence -- Periodicals; Expert systems (Computer science) -- Periodicals; Intelligent agents (Computer software) -- Periodicals; 006.33
• DOI: 10.1504/IJIIDS.2014.066630
• Languages: English
• ISSNs: 1751-5858

Abstract

Interlaced video acquisition and transmission, used as a simple way of increasing motion quality in video sequences, all by respecting the bandwidth constraints of progressive format representation, has been extensively used by analogue transmission systems. However, considering that nowadays the television sets and the flat screens can display only progressive format input, ensuring a smooth and high quality interlaced to progressive format conversion represents a challenging task. If till present, interpolation-based like solutions have been preferred for performing this conversion, in this paper we use an inverse problem formulation for video deinterlacing and propose a motion-compensated total-variation-minimisation-based reconstruction algorithm for solving it. A first recovery of the progressive frame is obtained using a contour-preserving interpolation method and further used for triggering a bidirectional motion-compensated prediction for the current field. A motion-compensated residual is calculated as difference between the current field and the projection of its temporal prediction using the same parity sampling matrix. This field residual is further reconstructed using a total-variation regularisation method and added back to the motion-compensated prediction to form the final progressive frame. As shown by the experimental results, the proposed deinterlacing method presents high quality progressive frame reconstructions compared to classical deinterlacing approaches.