FPGA Implementation of Multi-Spectral HDR Imaging

In this paper, we present a computationally efficient implementation of multi-spectral HDR (High Dynamic Range) image fusion on a Field Programmable Gate Array (FPGA), where an infrared HDR image stream is fused with a visible light HDR image stream. Each HDR frame in the video is composited from a set of differently exposed images of the same subject matter. This system runs in real-time and captures high-quality video for rugged industrial environments with difficult lighting conditions and fugitive dust/soot (e.g. in underground coal mining or arc welding). Firstly, the visible sensor is configured to stream video frames with alternated exposures. These differently exposed images are quantigraphically composited using an efficient implementation [1]. Then, the infrared image from a different sensor is synchronized and geometrically transformed to match the frames from the visible light sensor. Finally, the system fuses the HDR infrared video stream with the visible light HDR video stream to produce the output.