Time Complexity Optimization of HEVC Intra Coding Using Laplacian Edge detection in Video Compression

In contemporary video codecs, blocks exhibiting motion discontinuity undergo further partitioning into smaller units. This finer subdivision enables the acquisition of enhanced details compared to blocks with singular motion. Traditional codecs employ Rate Distortion Optimization (RDO) that is a technique to differentiate between blocks with motion discontinuity and those with solitary motion. However, this procedure demands a substantial investment of time and computational power, making it a resource-intensive process. Numerous research studies have proposed alternative models that utilize distinct features to determine the presence of motion discontinuity within a block of a video frame. In this research, a model is presented that can reduce the time complexity of intra-mode HEVC and also predict whether a block of a video frame contains motion discontinuity or not. Our method uses an edge detection model to find important edges inside a frame rather than depending on RDO. Edges that contain sensitive data are divided up to the maximum coding tree units (CTU) depth. We predict early CTU partition decisions in this method. Our experimental result shows that, the proposed model has saved an overall 44.67% encoding time complexity compared to HEVC standard with 3.25% bit rate increment as a compensation.