Please use this identifier to cite or link to this item:
https://scholarbank.nus.edu.sg/handle/10635/17375
Title: | Gradient variation: A key to enhancing photographs across illumination | Authors: | ZHANG XIAOPENG | Keywords: | Computational Photography, Illumination, NIR, Flash, Radiance Gradient, Image Enhancement | Issue Date: | 20-Jul-2009 | Citation: | ZHANG XIAOPENG (2009-07-20). Gradient variation: A key to enhancing photographs across illumination. ScholarBank@NUS Repository. | Abstract: | Extreme lighting conditions, such as a high dynamic range (HDR) scene or a low light environment, may introduce many kinds of defects for digital photography. These artifacts, including pixel saturation, noise, blurring, etc., have been studied as individual problems in the area of image processing for decades. Recently, Computational Photography, an emerging area across computer vision and computer graphics, attempts to resolve the problems by ¿reinventing¿ digital cameras. In light of previous contributions in this area, this thesis presents two novel solutions to enhance photographs in HDR scene and low light environment respectively. The key ideas behind both solutions are based on one fundamental observation: the image gradient variation due to illumination change. However, such variation has been either ignored or over simplified in the existing works. To enhance photography in HDR scenes, we build a dual-camera system, being able to capture a normal (visible-light) photograph and its near infra-red counterpart with a single shot. By manipulating the gradient magnitude based on near infra-red information, the visible photograph can be enhanced in both contrast and texture. For low light environment photograph enhancement, we propose a selective re-flashing method which requires only two input flash photographs with different flash intensities. By decomposing the radiance gradient, our method can faithfully recover the ambient image and allows the user to re-flash the selected regions. Compared to the existing works, our approach is simple yet efficient and able to achieve higher visual quality. Besides, we hope that our work can help better understand the relationship between gradient and illumination, and provide more insights for digital photograph enhancement. | URI: | http://scholarbank.nus.edu.sg/handle/10635/17375 |
Appears in Collections: | Ph.D Theses (Open) |
Show full item record
Files in This Item:
File | Description | Size | Format | Access Settings | Version | |
---|---|---|---|---|---|---|
thesis_final.pdf | 27.45 MB | Adobe PDF | OPEN | None | View/Download |
Google ScholarTM
Check
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.