The paper proposes a novel prediction paradigm in image coding based on Convolutional Neural Networks (CNN). A deep neural network is designed to provide accurate pixel-wise prediction based on a causal neighbourhood. The proposed CNN prediction method is trained on the high-activity areas in the image and it is incorporated in a lossless compression system for high-resolution photographic images. The system uses the proposed CNN-based prediction paradigm as well as LOCO-I, whereby the predictor selection is performed using a local entropy-based descriptor. The prediction errors are encoded using a CALIC-based reference codec. The experimental results show a good performance for the proposed prediction scheme compared to state-of-the-art predictors. To our knowledge, the paper is the first to introduce CNN-based prediction in image coding, and demonstrates the potential offered by machine learning methods in coding applications.
Schiopu, I, Liu, Y & Munteanu, A 2018, CNN-based Prediction for Lossless Coding of Photographic Images. in 2018 Picture Coding Symposium, PCS 2018 - Proceedings., 8456311, IEEE, San Francisco, CA, USA, pp. 16-20, 2018 Picture Coding Symposium, San Francisco, California, United States, 24/06/18. https://doi.org/10.1109/PCS.2018.8456311
Schiopu, I., Liu, Y., & Munteanu, A. (2018). CNN-based Prediction for Lossless Coding of Photographic Images. In 2018 Picture Coding Symposium, PCS 2018 - Proceedings (pp. 16-20). Article 8456311 IEEE. https://doi.org/10.1109/PCS.2018.8456311
@inproceedings{47c76eaecff5496fb52fbc376070f74c,
title = "CNN-based Prediction for Lossless Coding of Photographic Images",
abstract = "The paper proposes a novel prediction paradigm in image coding based on Convolutional Neural Networks (CNN). A deep neural network is designed to provide accurate pixel-wise prediction based on a causal neighbourhood. The proposed CNN prediction method is trained on the high-activity areas in the image and it is incorporated in a lossless compression system for high-resolution photographic images. The system uses the proposed CNN-based prediction paradigm as well as LOCO-I, whereby the predictor selection is performed using a local entropy-based descriptor. The prediction errors are encoded using a CALIC-based reference codec. The experimental results show a good performance for the proposed prediction scheme compared to state-of-the-art predictors. To our knowledge, the paper is the first to introduce CNN-based prediction in image coding, and demonstrates the potential offered by machine learning methods in coding applications.",
keywords = "Machine Learning, Prediction, lossless compression",
author = "Ionut Schiopu and Yu Liu and Adrian Munteanu",
year = "2018",
month = jun,
day = "25",
doi = "10.1109/PCS.2018.8456311",
language = "English",
isbn = "978-1-5386-4161-3",
pages = "16--20",
booktitle = "2018 Picture Coding Symposium, PCS 2018 - Proceedings",
publisher = "IEEE",
note = "2018 Picture Coding Symposium, PCS ; Conference date: 24-06-2018 Through 27-06-2018",
url = "http://www.pcs2018.com/",
}