Enhancing Versatile Video Coding Efficiency via Post-Processing of Decoded Frames Using Residual Network Integration in Deep Convolutional Neural Networks

The VVC standard, also known as Versatile Video Coding represents an advancement, in video compression technology. Its main objective is to improve compression efficiency while upholding high image quality standards. One of the standout features of VVC is its capacity to enhance performance in applications by reducing data requirements without compromising the contents integrity. This study presents a technique for enhancing video quality post compression by integrating a Residual Deep Convolutional Neural Network (Res DCNN) into the VVC framework. By combining the ResNet architecture with Deep Convolutional Neural Networks (DCNNs) the research aims to improve the clarity of video frames. The Res DCNN framework, comprising layers and Residual Dense Blocks (RDBs) effectively extracts features from video data. By using VVC Test Model 22.2 for encoding and decoding processes and TensorFlow for model training this study demonstrates improvements in video quality. The results indicate an enhancement in image quality with the PSNR value increasing from 35.36 dB to 37.33 dB highlighting the effectiveness of the Res DCNN approach. The seamless integration of Res DCNN into video compression technology marks a step forward in achieving visual quality while optimizing data usage, for video processing purposes.