Via introducing neural network theory and fuzzy sets theory , a novel neural network algorithm (NNBF) based on fuzzy sets is proposed. Image fuzzy sets are taken as feature space, and the exact positions are optimized by using 3 layers forward neural network algorithm. Experimental results show that the proposed algorithm is more advantagous in matching accuracy, speed, robustness, and has anti-jamming performance compared with conventional methods. It is fit for the real-time aircraft scene matching.

In order to solve the problem of ranging remote targets in IRST system, a method using triple-station IRST system is presented. Compared with the methods which use single-station of multiwave band IRST system and double-station either of single-wave band or multi-wave band IRST systems,this method possesses the merits of higher accuracy and more consistent in all azimuths. Simulation result shows that the IRST system can achieve a very high accuracy when the distance among the three stations is no less than 900 m. How to identify same targets among different stations in IRST system is also illustrated.

A new method to determine the optical constants and geometric thickness of single thin film is proposed with the measurement of the reflectance from spectrophotometer. Based on the Fresnel's optical expression, improved flexible tolerance method is employed in the case of a digital model of thin film by fitting the curve of measured reflectance spectrum. Simulation results show a satisfactory coherence of the optical constants and the thickness of the target film. By taking the influence of nonlinear condition into account with more direct and indirect limitation, the precision and value-searching efficiency have been improved. Furthermore, the problem of dimension degradation, which exists in "Downhill Simplex", has been successfully avoided. No initial input is needed for the procedure of optimization to achieve optical solution,which makes the whole processing of value calculation much more convenient and efficient.