Research and development of image improvement tools

The article deals with the problem of improving image quality using neural network methods. The goal of the research is to develop and study the effectiveness of software that improves image quality by using neural network technologies. The relevance of the work is due to the fact that the use of neural networks (NN) for the subject area under consideration allows the application of new methods that cannot be implemented by traditional approaches. This makes it possible to increase the efficiency and reduce the cost of image processing. The object of the study is graphic images. The subject of the study is the performance indicators (metrics) of neural network methods used to improve image quality. The paper analyzes traditional and neural network methods for increasing resolution and improving image quality. Among the traditional methods, the methods and algorithms of interpolation and filtering were analyzed, and among the neural network methods, the Super-Resolution, Inpainting and Denoising algorithms were analyzed. The paper also presents a comparative analysis of popular software solutions for increasing image resolution. The advantage of using NN methods is shown, and the relevance of software development is determined. The architectures of autoencoders, generative adversarial, convolutional and diffusion neural networks used to improve image quality are analyzed, their advantages and disadvantages are identified. The choice of NN architecture is substantiated. Criteria for assessing image quality are considered. The use of the Structural similarity index measure (SSIM) and Peak signal-to-noise ratio (PSNR) criteria is substantiated. A set of tools for software development is defined. The architecture of a neural network that will solve the problems of existing software solutions is developed. The analysis of the results of the proposed software operation is carried out and conclusions are made about its effectiveness. The work uses the method of theoretical analysis, the method of comparison, as well as experiment and generalization.