A feature based complexity model for decoder complexity optimized HEVC video encoding

Abstract

The complexity of the novel video compression algorithms is a major contributor for the increased demand of processing and energy resources for video playback in consumer electronic devices. Therefore, a decoder complexity reduction mechanism is proposed which constitutes a model that predicts the decoder's complexity requirements to decode the HEVC encoded bit streams with a 4.2% average prediction error and a decoder complexity optimized encoding algorithm, which reduces the decoding complexity by an average of 28.06% and 41.19% with a -1.91 dB and -2.46 dB impact to the BD-PSNR for the low delay P and random access configurations, respectively.