Conference material: "Proceedings of the International Conference on Computer Graphics and Vision “Graphicon” (19-21 September 2022, Ryazan)"
Authors:Kinev I.Å., Gebel G.V., Zhdanov D.D., Zhdanov A.D.
Comparison of Neural Network and Circular Filtering Algorithms for Synthesized RGB Images
A study of the causes of the conflict of vergence-accommodation of human vision in virtual and mixed reality systems has been conducted. Technical and algorithmic approaches to reduce and eliminate the conflict of vergence-accommodation in virtual reality systems are considered. As a technical solution, an approach was chosen that provides adaptive focusing of the eyepiece of a virtual reality system to the convergence point of a person's eyes, determined by the tracking system of his pupils. Possible algorithmic solutions providing focusing of the virtual reality image in accordance with the expected accommodation of human eyes are considered. As the main solutions, we consider the classical solution of image filtering in accordance with defocusing caused by natural accommodation at a given distance, and a solution in which the corresponding filtering is performed using neural network technologies. The advantages and disadvantages of the proposed solutions are considered. As a criterion of correctness, a visual comparison of the results of image defocusing with the solution obtained by physically correct rendering using a human eye model was used. The method of bidirectional stochastic ray tracing using backward photon maps was used as the basis for physically correct rendering. The paper presents an analysis of the advantages and disadvantages of the proposed solutions.