The paper considers the problem of enhancing image quality using neural networks with different types of multi-level distortions - contrast degradation, adding noise of various natures, image compression, etc. The widespread TID2013 database, which contains both original images and images modified using various types of distortions (25 basic images, 24 types of distortions, and 5 of their levels), was used as the image database for training neural networks. This database was divided into training (480 images), validation (360), and test (120) images. The images with the highest level of distortion, 5, were selected for enhancement. To study the problem of image quality enhancement, a modification of the MIRNet neural network and a neural network based only on convolutional layers were used. The features of the construction and parameters of neural networks, the structure of the convolutional neural network, loss and accuracy graphs in the learning process, and time parameters of learning are presented. The modification of the MIRNet network consisted of adjusting the procedure for forming training and validation images based on the TID2013 database, and changing some parameters of the neural network. The number of layers of the modified neural network was 1339 (including the input layer, Concatenate, Pooling2D, Reshape layers, etc.). The main layers were Conv2D convolutional layers. The mean square error was used as a measure of loss when training the neural network, and the peak signal-to-noise ratio and the mean absolute error were used as measures of accuracy. In the neural network based on convolutional layers, 7 convolutional layers with the number of channels of 128 and 256 were used (except for the last convolutional layer with 3 channels to obtain the resulting enhanced image). The mean absolute error for the training images was 0,0491, and for the validation, 0,0527. Convolutional layers with the same number of channels in each layer were used. In this case, no overfitting effect was observed. When using an irregular structure of convolutional layers (for example, with 32 and 128 channels), the phenomenon of overfitting was observed. When conducting computer experiments on training of neural networks, an NVIDIA GeForce RTX 4080 graphics adapter was used.
[1] Kaur, A., and Gagandeep (2018), “A Review on Image Enhancement with Deep Learning Approach”, ACCENTS Transactions on Image Processing and Computer Vision, vol. 4 (11), pp. 16-20, doi: 10.19101/TIPCV.2018.411002
[2] Vu, T., Nguyen, C. V., Pham, T. X., Luu, T. M., and Yoo, C. D. (2019), “Fast and Efficient Image Quality Enhancement via Desubpixel Convolutional Neural Networks”, LNCS, Springer, Cham, vol. 11133, pp. 243-259, doi: 10.1007/978-3-030-11021-5_16
[3] Chen, X. (2019), “Image Enhancement Effect on the Performance of Convolutional Neural Networks”, Thesis for the degree of Master of Science in Computer Science, Blekinge Institute of Technology, Sweden, 40 pp. Available: https://www.diva-portal.org/smash/get/diva2:1341096/FULLTEXT02.pdf (accessed 08.05.2025)
[4] Kinoshita, Y., and Kiya, H. (2019), “Convolutional Neural Networks Considering Local and Global Features for Image Enhancement”. Available: https://arxiv.org/pdf/1905.02899, 5 pp. (accessed 08.05.2025), doi: 10.48550/arXiv.1905.02899
[5] https://keras.io/ (accessed 08.05.2025)
[6] Convolutional Autoencoder for Image Denoising. Available: https://keras.io/examples/vision/autoencoder/ (accessed 08.05.2025)
[7] Image Super-Resolution Using an Efficient Sub-Pixel CNN. Available: https://keras.io/examples/vision/super_resolution_sub_pixel/ (accessed 08.05.2025)
[8] Low-Light Image Enhancement Using MIRNet. Available: https://keras.io/examples/vision/mirnet/ (accessed 08.05.2025)
[9] Enhanced Deep Residual Networks for Single-image Super-Resolution. Available: https://keras.io/examples/vision/edsr/ (accessed 08.05.2025)
[10] Zero-DCE for Low-Light Image Enhancement. Available: https://keras.io/examples/vision/zero_dce/ (accessed 08.05.2025)
[11] Shi, W., Caballero, J., Huszar, F., Totz, J., Aitken, A. P., Bishop, R., Rueckert, D., and Wang, Z. (2016), “Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network”. Available: https://arxiv.org/pdf/1609.05158, 10 pp. (accessed 08.05.2025), doi: 10.48550/arXiv.1609.05158
[12] Zamir, S. W., Arora, A., Khan, S., Hayat, M., Khan, F. S., Yang, M.-H., and Shao, L. (2020), “Learning Enriched Features for Real Image Restoration and Enhancement”, Available: https://arxiv.org/pdf/2003.06792, 20 pp. (accessed 08.05.2025), doi: 10.48550/arXiv.2003.06792
[13] Lim, B., Son, S., Kim, H., Nah, S., and Lee, K. M. (2017), “Enhanced Deep Residual Networks for Single Image Super-Resolution”, Available: https://arxiv.org/pdf/1707.02921, 9 pp. (accessed 08.05.2025), doi: 10.48550/arXiv.1707.02921
[14] Guo, C., Li, C., Guo, J., Loy, C. C., Hou, J., Kwong, S., and Cong, R. (2020), “Zero-Reference Deep Curve Estimation for Low-Light Image Enhancement”, Available: https://arxiv.org/pdf/2001.06826, 10 pp. (accessed 08.05.2025), doi: 10.48550/arXiv.2001.06826
[15] Ponomarenko, N., Jin, L., Ieremeiev, O., Lukin, V., Egiazarian, K., Astola, J., Vozel, B., Chehdi, K., Carli, M., Battisti, F., and Kuo, C.-C. J. (2015), “Image database TID2013: Peculiarities, results and perspectives”, Signal Processing: Image Communication, vol. 30, January, pp. 57-77, doi: 10.1016/j.image.2014.10.009
[16] Ponomarenko, N., Ieremeiev, O., Lukin, V., Jin, L., Egiazarian, K., Astola, J., Vozel, B., Chehdi, K., Carli, M., Battisti, F., and Kuo, C.-C. J. (2013), “A new color image database TID2013: Innovations and results”, Proceedings of ACIVS, Poznan, Poland, Oct., pp. 402-413, doi: 10.1007/978-3-319-02895-8_36
[17] Chollet, F. Deep Learning with Python, 2nd ed., Manning Publications Co., 2021, 478 pp.