PhD Thesis Defense

Title: High Dynamic Range Video Merging, Tone Mapping, and Real-time Implementation

By: Mr. Willie C Kiser

Advisor: Dr. Ramiro Jordan

Date: Aug 10th 2012, 10:00 AM

Location: ECE, Room 118

Although high dynamic range (HDR) imaging has been the subject of significant research over the past fifteen years, the goal of cinema-quality HDR video has not yet been achieved. In this work, I reference an optical method patented by Contrast Optical which is used to capture sequences of Low Dynamic Range (LDR) images that can be used to form HDR images as the basis for HDR video. Because of the large difference in exposure spacing of the LDR images captured by this camera, present methods of merging LDR images are insufficient to produce cinema quality HDR images and video without significant visible artifacts. Thus, the focus of my research presented is two-fold, first a new method of combining LDR images with exposure differences of greater than 3 stops into an HDR image is presented. Second a method of tone mapping video based on the Reinhard photographic operator is presented which solves potential problems of HDR video flicker and automated parameter control of the tone mapper itself. A prototype and subsequent production of this video capture system along with the combining and tone mapping algorithms have been implemented in a high-definition HDR-video system. Additionally, FPGA hardware implementation details are given to support real time HDR video. I will present still frames from the acquired HDR video combined and tone mapped with these techniques. Video results are also available and will be shown.