Decoder-Side Quality Enhancement of JPEG Images Using Deep Learning-Based Prediction Models for Quantized DCT Coefficients

• Antonio José G. Busson PUC-Rio
• Paulo Renato C. Mendes PUC-Rio
• Daniel de S. Moraes PUC-Rio
• Álvaro Mário G. da Veiga PUC-Rio
• Sérgio Colcher PUC-Rio
• Álan Lívio V. Guedes PUC-Rio

Resumo

Many recent works have successfully applied some types of Convolutional Neural Networks (CNNs) to reduce the noticeable distortion resulting from the lossy JPEG compression technique. Most of them are built upon the processing made on the spatial domain. In this work, we propose a JPEG image decoder that is purely based on the frequency-to-frequency domain: it reads the quantized DCT coefficients received from a low-quality image bitstream and, using a deep learning-based model, predicts the missing coefficients in order to recompose the same image with enhanced quality. In experiments with two datasets, our best model was able to improve from images with quantized DCT coefficients corresponding to a Qualityz Factor (QF) of 10 to enhanced quality images with QF slightly higher than 20.

Palavras-chave: Deep learning, Convolutional Neural Networks, Image compression, JPEG, DCT