Neural reflectance transformation imaging
| Autor: | Federico Ponchio, Filippo Andrea Fanni, Fabio Pellacini, Andrea Giachetti, Tinsae Gebrechristos Dulecha |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2020 |
| Předmět: |
Relighting · Neural network
Computer science ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Relighting 02 engineering and technology Benchmark Rendering (computer graphics) Computer graphics 0202 electrical engineering electronic engineering information engineering Computer vision Interactive visualization ComputingMethodologies_COMPUTERGRAPHICS Pixel business.industry 020207 software engineering Autoencoder Neural network Reflectance transformation imaging Computer Graphics and Computer-Aided Design Data exchange Virtual image 020201 artificial intelligence & image processing Computer Vision and Pattern Recognition Artificial intelligence Polynomial texture mapping business Reflectance transformation imaging Relighting · Neural network Autoencoder Benchmark Software |
| Zdroj: | The visual computer (2020): 2161–2174. doi:10.1007/s00371-020-01910-9 info:cnr-pdr/source/autori:Dulecha T. G.; Fanni F. A.; Ponchio F.; Pellacini F.; Giachetti A./titolo:Neural reflectance transformation imaging/doi:10.1007%2Fs00371-020-01910-9/rivista:The visual computer/anno:2020/pagina_da:2161/pagina_a:2174/intervallo_pagine:2161–2174/volume |
| DOI: | 10.1007/s00371-020-01910-9 |
| Popis: | Reflectance transformation imaging (RTI) is a computational photography technique widely used in the cultural heritage and material science domains to characterize relieved surfaces. It basically consists of capturing multiple images from a fixed viewpoint with varying lights. Handling the potentially huge amount of information stored in an RTI acquisition that consists typically of 50–100 RGB values per pixel, allowing data exchange, interactive visualization, and material analysis, is not easy. The solution used in practical applications consists of creating “relightable images” by approximating the pixel information with a function of the light direction, encoded with a small number of parameters. This encoding allows the estimation of images relighted from novel, arbitrary lights, with a quality that, however, is not always satisfactory. In this paper, we present NeuralRTI, a framework for pixel-based encoding and relighting of RTI data. Using a simple autoencoder architecture, we show that it is possible to obtain a highly compressed representation that better preserves the original information and provides increased quality of virtual images relighted from novel directions, especially in the case of challenging glossy materials. We also address the problem of validating the relight quality on different surfaces, proposing a specific benchmark, SynthRTI, including image collections synthetically created with physical-based rendering and featuring objects with different materials and geometric complexity. On this dataset and as well on a collection of real acquisitions performed on heterogeneous surfaces, we demonstrate the advantages of the proposed relightable image encoding. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|