Optimal Cost Modeling Scheme for Efficient Intra-Prediction Mode in Video Compression

Abstract

It can be seen that the success rate of the video compression standard H.264 is quite high owing to its wide range of adoption. However, the legacy version of video compression techniques built on the top of H.264 might ensure reduced bitrate while preserving the quality of the reconstructed video signal to some extent, but it is achieved at the price of compression complexity. Our investigation reveals that H.264 achieves higher compression efficiency with MPEG files but does not ensure optimized performance for intra-prediction mode decision with rate distortion. The prime reason lies into less emphasis on redesigning the encoder with optimized cost function which could improvise the performance of existing Lagrangian cost function (LCF). This area of research was explored and accentuated in many studies but still very few talked in a favor to comprehend the cost optimization performance for encoding I-frame and P-frame in MPEG videos. Unlike existing literature, the study found scope of redesigning the H.264-based encoder module considering the baseline of LCF and in addition introduces a novel cost function to optimize the flow of execution of encoding. In this regard, a computational framework is introduced for realizing the numerical modeling integrating both LCF and proposed CF for H.264 intra-prediction mode decision. The novelty of this system is that it does not have dependency on the operations of entropy coding during rate-distortion (RD) optimization and this approach positively impacts on the computing performance of the system. The evaluated performance outcome shows a detailed comparison between both the methods on the basis of peak-signal-to-noise ratio (PSNR), bit-error rate (bit/s) and encoding time (s). Unlike LCF, the outcome of the study claims its effectiveness in maintaining a well-balanced performance between reconstructed signal quality and encoding time.