For military infrared devices, whether the required scene information can be captured is the key front-end task, and whether the effective detection, recognition, tracking and intelligence of the target (scene) can be completed is the image processing of the latter part. The subsystem completes by further processing the previous signal.

(1) Target (scene) detection The target information that the infrared imaging system can detect at a long distance is usually very weak, the observed target image is often point-like, and the target is often in various motion states in the field of view. However, the backgrounds where the targets are located are very different. Therefore, the detection of weak and small moving targets in complex backgrounds has become a research hotspot.

Modern warfare requires infrared detection systems to find and track targets from a long distance, so as to gain time for command system decision-making and weapon systems. The infrared detection system works in a passive way, with strong anti-interference ability, good concealment, but short operating distance. Because the spatial resolution of the optical system has reached or is close to the theoretical limit level. The more practical method is to make up for the short range of the infrared detection system by improving the performance of the target detection algorithm, especially the detection performance of weak and small targets. "Weak" and "small" refer to two aspects of the target's properties. The so-called "weak" refers to the infrared radiation intensity of the target, which is reflected on the image and refers to the grayscale of the target; the so-called "small" refers to the size of the target, reflecting the To the image is the number of pixels occupied by the indicator. Infrared weak and small target detection and identification difficulties are: low contrast, blurred edges, weak signal strength, lack of texture, shape, size and other structural information, the target is easily overwhelmed by noise, single-frame detection has a high false alarm rate, and multi-frame processing adds The amount of data storage and calculation, fixed templates and operators are difficult to effectively detect weak and small targets.

Target detection algorithms in actual combat situations generally include morphological algorithms, noise filtering (wavelet transform, Contourlet transform, etc.) algorithms, multi-frame accumulation statistical algorithms, and adaptive background prediction algorithms. The effectiveness of algorithms is often only suitable for specific application scenarios.

(2) Self-identification should firstly be able to detect the shape characteristics and radiation characteristics of various targets and interferers, and determine their respective motion trajectories, and then use the identification algorithm to automatically identify and determine the target.

(3) Accurate target tracking must first determine the orientation of the target and its motion state. For imaging targets, it is necessary to determine the target orientation based on the target's centroid, centroid or key parts of the target (such as cockpit, fuel tank, warhead).

It can be seen that the amount of calculation required for military infrared image processing is very large, and the calculation speed is required to be very high. The current research progress of high-speed and large-capacity computers provides a realistic possibility for the development of military infrared imaging devices, and the development efficiency can be greatly improved due to the use of FPGA+DSP.

With the development of large-scale programmable devices, the signal processing system using DSP+FPGA structure has shown its superiority and is being gradually paid attention to. Field Programmable Gate Array (FPGA) is developed on the basis of special-purpose integrated circuit chips, which overcomes the inflexibility of special-purpose integrated circuits. Its advantage is that it has strong flexibility, that is, its internal specific logic functions can be configured as needed, and it is very convenient to modify and maintain the circuit. At present, the capacity of FPGA has reached millions of gates, making FPGA a An important option for addressing system-level designs.