Real-time HDR video imaging on FPGA with compressed comparametric lookup tables

• Published in: 2014 IEEE 27th Canadian Conference on Electrical and Computer Engineering (CCECE) pp. 1 - 6
• Main Authors: Tao Ai, Ali, Mir Adnan, Steffan, Greg, Ovtcharov, Kalin, Zulfiqar, Sarmad, Mann, Steve
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.05.2014
• Subjects: Cameras; Field programmable gate arrays; Interpolation; Pipeline processing; Streaming media; Table lookup
• ISBN: 1479930997; 9781479930999
• ISSN: 0840-7789
• DOI: 10.1109/CCECE.2014.6901124

Video cameras can only take photographs with limited dynamic range. One method to overcome this is to combine differently exposed images of the same subject matter (i.e. a Wyckoff Set), producing a High Dynamic Range (HDR) result. Implementations for real-time HDR videos have relied upon methods that are less accurate. Instead of weighted-sum approaches that adds up noise for large number of LDR images, lookup tables that contains results calculated using a probabilistic model can be generated. This lookup table, when compressed using quadtree structure, can be implemented on a medium-sized FPGA. The work presented in this paper improves over the earlier publication to support high-definition videos. To achieve the required bandwidth for HDR composition, the evaluation of the camera response function is performed within a compressed lookup table that has its address retrieved using a pipelined multiplexer network. We are able to generate the entire circuit using software. It is parameterizable by user-specified error constraints, allowing us to explore the trade-offs in resource usage and precision of the implementation.