Approximate-Computing Architectures for Motion Estimation in HEVC
Alberto Paltrinieri, Riccardo Peloso, Guido Masera, Muhammad Shafique, Maurizio Martina · 2018
Video compression deeply relies on motion estimation but this crucial step requires power-hungry computational resources. Starting from an existing architecture capable of calculating a high number of Sum of Absolute Differences (SADs), three possible locations for approximate adders are chosen. At each location, different types of approximate adders are implemented and the results are analyzed in terms of error and power saving.