Non-Iterative Tone Mapping With High Efficiency and Robustness

• Published in: IEEE access 2018-01, Vol.6, p.35720-35733
• Main Authors: Bae, Gyujin, Jang, Chan Young, Cho, Sung In, Kim, Young Hwan
• Format: Article
• Language: English
• Subjects: Benchmark testing; Benchmarks; Computational complexity; guided filter; high-dynamic range compression; Image coding; Image edge detection; Image quality; Indexes; Iterative methods; Lighting; Luminance; Mapping; Methods; Optimization; Performance evaluation; statistical model; structure fidelity; Tone mapping
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2018.2846772

This paper proposes an efficient approach for tone mapping, which provides a high perceptual image quality for diverse scenes. Most existing methods, optimizing images for the perceptual model, use an iterative process and this process is time consuming. To solve this problem, we proposed a new layer-based non-iterative approach to finding an optimal detail layer for generating a tone-mapped image. The proposed method consists of the following three steps. First, an image is decomposed into a base layer and a detail layer to separate the illumination and detail components. Next, the base layer is globally compressed by applying the statistical naturalness model based on the statistics of the luminance and contrast in the natural scenes. The detail layer is locally optimized based on the structure fidelity measure, representing the degree of local structural detail preservation. Finally, the proposed method constructs the final tone-mapped image by combining the resultant layers. The performance evaluation reveals that the proposed method outperforms the benchmarking methods for almost all the benchmarking test images. Specifically, the proposed method improves an average tone mapping quality index-II (TMQI-II), a feature similarity index for tone-mapped images (FSITM), and a high-dynamic range-visible difference predictor (HDR-VDP)-2.2 by up to 0.651 (223.4%), 0.088 (11.5%), and 10.371 (25.2%), respectively, compared with the benchmarking methods, whereas it improves the processing speed by over 2611 times. Furthermore, the proposed method decreases the standard deviations of TMQI-II, FSITM, and HDR-VDP-2.2, and processing time by up to 81.4%, 18.9%, 12.6%, and 99.9%, respectively, when compared with the benchmarking methods.