AVS Intra Prediction Algorithm Implementation

In this project, we implemented the AVS intra prediction algorithm using MATLAB code that specifically handles four prediction modes: horizontal prediction, vertical prediction, DC/top-left prediction, and bottom-right prediction. The implementation includes comprehensive explanations and demonstrations of these prediction methods, with code comments detailing the algorithmic approach for each mode. For horizontal prediction, the algorithm extrapolates pixel values from left neighboring samples using vectorized operations, while vertical prediction replicates values from above samples through matrix manipulations. The DC prediction calculates average values from available reference samples, and the directional predictions employ angle-based interpolation methods. Our analysis covers both the advantages and limitations of each prediction method, including computational efficiency comparisons between different implementations. Additionally, we explored optimization techniques such as pre-computation of reference samples and memory-efficient data structures to enhance algorithm performance. The project not only delivers a functional AVS intra prediction implementation but also provides in-depth technical insights into prediction methodology details, code optimization strategies, and potential applications in video coding systems.