Fast AVS to HEVC Transcoding via Visual Feature Extraction with Resolution Conversion

Video transcoding is the important technique and is widely used in many multimedia applications. In the proposed method, we mapped the Advance Video Coding Standard (AVS) to High Efficiency Video Coding Standard (HEVC) using Super Resolution (SR) Technique. To improve transcoding with better visual efficiency, we developed a technique based on SR using feature extraction. Implementing this approach, we have improved resolution for AVS video bitstream to High Definition (HD) video bitstream. This method has three kinds of feature extraction, Transform Coefficients (TC), Macro Blocks (MB) modes (Intra/Inter) and decoded Motion Vectors (MV). The developed method classifies each video frame into two different type of regions: Most Region of Interest (MROI) and Less Region of Interest (LROI). Afterward, SR technique is applied on decoded low resolution regions with different parameter values to get high resolution regions. The experimental results show that the proposed method achieves an average increase of 0.07% dB in PSNR with similar SSIM index with negligible bit rate increment.