Edge-type High-speed Intelligent Infrared Pre-processor

This paper proposes an edge-type high-speed intelligent infrared pre-processor, including the pre-processing algorithm and the corresponding hardware architecture. At algorithm level, guided filtering method and neural network method are combined to propose a non-uniformity correction algorithm with stronger scene adaptability. By combining unsharpen mask and adaptive gray stretching algorithms, the processing effect of image details is effectively improved. At hardware level, this paper optimizes the above algorithm. The structure adopts modular pipeline design, which can realize dynamic switching between algorithms through control instructions, and can effectively meet the real-time pre-processing requirements of infrared images at edge end.