Fast video compression

Abstract

Uncompressed video data contains a significant amount of redundancy which can be exploited to reduce the amount of data required to store or transmit it. The latest video compression standard, High Efficiency Video Coding (HEVC) was developed by the ITU-T Video Coding Experts Group (VCEG) and the ISO/IEC Moving Picture Experts Group (MPEG) as a partnership known as the Joint Collaborative Team on Video Coding (JCTVC). The standard is designed on the same structure of the previous standard called H.264/AVC (Advanced Video Coding) but has incorporated multiple improvements such as decreased bit rate and compression efficiency but at the cost of increased complexity, due to the increased number of predictions and searches that need to be performed. The reference codec, called the HEVC test model (HM), is provided as a common reference implementation of the HEVC standard but is only provided to demonstrate the functionality of the codec and is not particularly optimised for speed or efficiency. To speed up the encoding process of the reference codec, functions consuming a large part of the encoding time are modified to utilise the entire CPU through multi-threading provided by the OpenMP API. The functions consuming the most of the encoding time are identified by profiling the unmodified HM encoder. Profiling results give various performance metrics that help to identify where the most work is being done by the encoder. Parallelisation is implemented into the HM encoder based on the results obtained from the profiling stage. The unmodified and modified encoders are compared with each other by encoding various video sequences using the same encoder configuration throughout. The implemented parallelisation resulted in a reduction of encoding time from ⇠10.5% to ⇠22.6% in the performed tests, with a negligible decrease in video quality of up to ⇠0.7% and a small increase in bit rate of up to ⇠4.4%.