Proceedings of The 11th International Conference on Research in Engineering, Science & Technology
Year: 2021
DOI: https://www.doi.org/10.33422/11th.restconf.2021.03.25
Analysis of Computational Vision Algorithms for Generation of Textures for Realistic Rendering
Carlos Eduardo Capanema; Gustavo Aquino Torres Teixeira, Luiz Melk de Carvalho, Diva de Souza e Silva Rodrigues, Flávio Henrique Batista de Souza
ABSTRACT:
This research presents an application based on the combination of image editing algorithms for the generation of auxiliary textures. It focuses on the generation of Specular Map and Bump Map from a Diffuse texture, to enable the automatic generation of high-quality images for games and movies. The image database for experiments was separated into two groups: Regular and Irregular, with 21 images of each and with different resolutions. The comparison solution between the algorithms was developed in C #. In the proposed software, it is possible to visualize the result of the texture processing in real time on a window where the 3D rendering occurs, using the Unity graphic engine, with the texture applied in a plane. For the manipulation of textures, instead of the standard implementation (pixels are obtained and changed by calling the Bitmap GetPixel and SetPixel functions), the manipulation takes place directly in an Array. The experiment had multiple analyzes of the following algorithms combined in pairs: Histogram Equalization, Adaptive Histogram Equalization, Contrast Limited Histogram Equalization, Grayscale Equal Weights, Desaturation, Grayscale with Luma Correction. The evaluated parameters were the time to generate the textures Bump Map and Specular Map, as well as the minimum and maximum values of parameters outputted by the algorithms. Those parameters represent a visual qualitative analysis of the final render. In preliminary tests the proposed strategy proved to be at least 4 and up to 70 times faster than the standard literature’s version for adjusting brightness, grayscale and contrast in a 1080 x 768 image resolution.
Keywords: Image processing; Rendering process; Specular Map; Bump Map; Diffuse.