An Energy Efficient JPEG Encoder with Neural Network Based Approximation and Near-Threshold Computing

Abstract

JPEG compression is an important part in low-power multimedia applications. This paper proposes an approach that leverages the error resilience of JPEG for different energy budgets. We select and train neural networks to approximate DCT and quantization code regions in JPEG. Then we design an architecture called reconfigurable neural unit (RNU) to accelerate trained neural networks which replace original codes. In addition, some architecture innovations are proposed to make our JPEG encoder works efficiently in near-threshold voltage region. This JPEG encoder synthesized with a 40nm CMOS technology, is able to operate at 40MHz for a 0.6V supply voltage. Results show up to 5.0 × energy reduction with 2.5 × performance degradation when compared to using a 1.0V nominal supply voltage.