MULTIPLE OBJECTIVE FRAME RATE UP CONVERSION (WedAmPO1)

Author(s) : Tak-Song Chong (Hong Kong University of Science and Technology, Hong Kong) Oscar C. Au (Hong Kong University of Science and Technology, Hong Kong) Wing-San Chau (Hong Kong University of Science and Technology, Hong Kong) Tai-Wai Chan (Hong Kong University of Science and Technology, Hong Kong)

Abstract : In this paper, we propose a multiple objective frame rate up conversion algorithm (MOFRUC), which utilizes two different models. The first model is a constant velocity model that assumes the objects position is a linear function of time. The second model exploits the spatial correlation between neighboring blocks, and assumes the pixel intensity is highly correlated in a small local area. In this model, the perceptual quality of interpolated frame is also taken into account and the blocking artifact is minimized. Our proposed MOFRUC estimates the motion trajectory by the first model and interpolate the frame along the motion trajectory. At the same time, the algorithm refines the motion trajectory by maximizing a spatial correlation measurement defined in the second model and interpolates the frame with minimum blocking artifact. In our simulation, the proposed MOFRUC achieves the objectives in both models simultaneously, and produces high quality interpolated frame.

Menu