Research On Clear Imaging Technology Of Moving Target In Photoelectric Tracking System

The photoelectric tracking system is a long-distance optical tracking imaging system that integrates light,machine and electricity.In recent years,it has been used as a supplementary equipment for radar detection in marine search,range measurement,national defense security and other fields.It has an irreplaceable role in imaging.The system adopts closed-loop control,the target information collected by the pointer is processed by the image processing board and then transmitted to the display console,and then the display console generates instructions according to the image information to control the pointer to realize the monitoring and tracking of the target.However,due to the moving target,the system imaging will have different degrees of defocus blur,which affects the tracking accuracy of the photoelectric tracking system to the target and the clarity of forensics.Therefore,based on the photoelectric tracking system,this article will carry out in-depth research on the clear imaging technology of moving targets.First,the hardware selection of the system platform is analyzed and discussed.Based on the ?COS-? embedded operating system and Quartus ? integrated development environment,the software design of the control board and image processing board is implemented,and the system control zoom,video acquisition,transmission,decoding and display.At the same time,the theoretical basis that affects the clear imaging of the optical system is discussed.Secondly,the image clarity evaluation function based on spatial domain,frequency domain,gray level entropy and statistics is analyzed.In view of the shortcomings,an adaptive double-filter evaluation function based on Lo G operator is proposed.Based on the traditional Lo G operator,this method introduces mask templates constructed with different response factors,and implements adaptive template selection according to the picture collection situation,which improves the evaluation accuracy of the function.At the same time,the improved median filter is added to remove the pepper and salt noise of the image,the noise point is determined by setting a threshold,and the noise point is subjected to a median filter processing with a mask size of 3×3.After testing,the improved median filtering takes less time and the edge retention effect is more complete.Then,the construction algorithm of dynamic focus window is studied.The effects of different focus window sizes and different focus areas on the value of the sharpness evaluation function are discussed through experiments.Based on the analysis of the limitations of the traditional focus window construction algorithm,a multi-scale focus window construction algorithm based on morphology is proposed.This method extracts multi-scale feature maps by performing pyramid processing on the image,combined with morphological erosion and dilation processing to denoise and expand the target area of the binarized feature maps,effectively achieving the construction of the focus window.Finally,combined with the definition evaluation function proposed in this paper and the construction algorithm of the dynamic focus window,a set of clear imaging search strategies is designed for the photoelectric tracking system.In terms of evaluation accuracy,noise immunity and stability,the sharpness evaluation function proposed in this paper is compared with the traditional evaluation function.The experimental results show that the evaluation function of this paper has excellent performance in terms of sharpness ratio,steepness and sensitivity.At the same time,the design experiment verifies the window construction effect of the dynamic focus window construction algorithm proposed in this paper under different degrees of defocus,and compares the moving target in the construction of the focus window with the traditional algorithm.The experimental results show that the algorithm in the focus window the construction effect has been greatly improved,and the effective window construction rate reaches 95%,which is much higher than other focus window algorithms,and meets the requirements of the photoelectric tracking system for the focus window construction.