Localized high dynamic range plenoptic image compression

Chuan Chung Chang, Hsin Hsiang Lo, Han Hsuan Lin, Zhi Rong Fan, Shao Hsuan Cheng, Chih Hung Lu, Fu Ming Chuang, Jiun-In Guo

研究成果: Conference article同行評審

摘要

Based on multi-view characteristics of p lenoptic imaging, a new method for generating lo calized h igh dynamic range (LHDR) compression for plenoptic image is proposed, which determine HDR compression region, and maintains image quality for different plenoptic-views. The proposed method is realized by physically segmented sub-images through plenoptic optics. And plenoptic image with wider dynamic range is processed by calcula ted lo cal contrast of combined sub-images via block based harsh environment detection, and then followed by HDR process or dynamic local contrast enhancement (DLCE) for each block. Experiment results for proposed LHDR and reference HDR are also reported, which using self-designed p lenoptic camera with low distorted sub-image, 60° field of view, 1920×1080 pixels, and 90 fps 2/3″ CMOS imager. The results show that the proposed method is capable of wider dynamic range compression, while preserving fine details, fewer artifacts. Ext ended d ynamic range from 84.6 dB to 108.9 dB is realized. And processing time of proposed method is 85% compared to reference HDR processing.

原文English
頁(從 - 到)163-168
頁數6
期刊IS and T International Symposium on Electronic Imaging Science and Technology
DOIs
出版狀態Published - 1 1月 2017
事件Computational Imaging XV 2017 - Burlingame, United States
持續時間: 29 1月 20172 2月 2017

指紋

深入研究「Localized high dynamic range plenoptic image compression」主題。共同形成了獨特的指紋。

引用此