Low power motion estimation based frame up-conversion hardware designs
Özcan, Tevfik Zafer (2012) Low power motion estimation based frame up-conversion hardware designs. [Thesis]
Recently flat panel high definition television (HDTV) displays with 100 Hz, 120 Hz and 240 Hz picture rates are introduced. However, video materials are captured and broadcast in different temporal resolutions ranging from 24 Hz to 60 Hz. In order to display these video formats correctly on high picture rate displays, new frames should be generated and inserted into the original video sequence to increase its frame rate. Therefore, frame rate upconversion (FRUC) has become a necessity. Motion compensated FRUC (MC-FRUC) algorithms provide better quality results than non-motion compensated FRUC algorithms. These MC-FRUC algorithms consist of two main stages, motion estimation (ME) and motion compensated interpolation (MCI). In ME, motion vectors (MV) are calculated between successive frames, and in MCI this MV data is used to generate a new frame that is inserted between two successive frames, thus doubling the frame rate. In addition to these two main steps, intermediate steps such as refinement of the MV field by various algorithms like motion vector smoothing and bilateral ME refinement may be used to improve the quality of the interpolated video. In this thesis, a perfect absolute difference technique for block matching ME hardware is proposed. The proposed technique reduces the power consumption of a full search ME hardware by 2.2% on a XC2VP30-7 FPGA without any PSNR loss. In addition, a global motion estimation (GME) algorithm and its hardware implementation are proposed. The proposed GME algorithm increases PSNR of 3D recursive search ME algorithm by 2.5% and its hardware implementation is capable of processing 341 720p frames per second. An adaptive technique for GME, which reduces the energy consumption of the GME hardware by 14.37% on a XC6VLX75T FPGA with a 0.17% PSNR loss, is also proposed. Furthermore, an early termination technique for the adaptive bilateral motion estimation (ABIME) algorithm is proposed. The proposed technique reduces the energy consumption of the ABIME hardware by 29% with a 0.04% PSNR loss on a XC6VLX75T FPGA. In addition, an efficient weighted coefficient overlapped block motion compensation (WC-OBMC) hardware which reduces the dynamic power consumption of the reference WC-OBMC hardware by 22% is proposed. The proposed hardware is capable of processing 57 720p frames per second on a XC6VLX75T FPGA. Finally, the ABIME hardware is implemented on a Xilinx ML605 FPGA board.
Repository Staff Only: item control page