A low delay highly dynamic range infrared imaging system for complex scenes based on FPGA

We proposed a hardware-oriented framework for infrared images compression and enhancement. The aim is to achieve high-quality imaging with minimal delay. Specifically, we introduce several key improvements: Firstly, we adopt the anisotropic guided filtering for layering, which effectively addresses the issue of detail halo. Secondly, we proposed a dynamic range compression algorithm (LIHS&HRC), which is based on local information histogram statistics and thus highlights the gray scales with abundant details. Thirdly, we propose a low-complexity adaptive two-dimensional non-local mean algorithm, which preserves important details while reducing noise. Through extensive experimentation, it is demonstrated the proposed algorithm framework significantly reduces latency and Field Programmable Gate Array (FPGA) resource requirements. It generates high-quality infrared image sequences in a minimal delay of only 0.31 ms at a frame rate of 50fps, with consuming a mere 0.0362 MB of shift register and 0.0312 MB of BRAM.