Method and architecture design for motion compensated frame interpolation in high-definition video processing

Digital displays such as Liquid Crystal Display (LCD) and plasma display televisions have become prevalent in recent years. Sports broadcasting and movies are two prime factors responsible for this popularity. However, motion blur and judder appear as objects move rapidly or color dramatically changes on a wide range of LCD devices because of slow response time and sample-and-hold drive nature. Frame Rate Up Conversion (FRUC) is a well-studied method that is used to minimize these detrimental effects. A novel, fast, and efficient method with a well-designed architecture is proposed for Motion Compensated Frame Interpolation (MCFI) or Frame Rate Up Conversion. Unlike previous works involving high complexity, time-consuming iterations, and higher complex architecture, the proposed method adopts a one-pass, low-complexity approach without any iteration and is capable of dealing with High Definition (HD) video processing. Rather than a conventional motion estimation in MCFI, our method employs a unique true motion engine that explores at most nine motion candidates with different motion directions and then determines one true motion by referring to neighboring spatial and temporal information. For the purpose of motion estimation, the proposed method introduces an adaptive overlapped block matching algorithm known as the Multi-Directional Enlarged Matching Algorithm (MDEMA), and considers different overlapped types based on the direction of the current motion vector in order to enhance searching accuracy and visual quality. For practical issues and real-time HD requirements, the proposed architecture employs a modified Multi-level Successive Eliminate Algorithm (MSEA), which is a Fast Full-Search Block Matching Algorithm (FFSBMA) and has the ability to reduce the heavy computation of full search while maintaining similar quality. According to analyzed temporal information, our method explores true motion candidates and refines the accuracy of true motions for blocks or sub-blocks. Experimental results show that the proposed algorithm provides better video quality than conventional methods and demonstrates excellent performance for 30fps HD1080p video (1920x1080 resolution) at 180MHz or 30fps 720p video (1280x720) at 83MH