Design Of Division-of-Focal-Plane Polarization Imaging System And Image Fusion Method

Target detection in complex environments is the key to winning modern warfare.Traditional optical imaging systems detect targets through light intensity information,but when the light intensity information is disturbed and greatly attenuated,traditional optical imaging systems cannot effectively detect targets.The polarization imaging system combines polarization imaging technology on the basis of the traditional optical imaging system,which can not only collect light intensity information,but also collect polarization information.The polarization information is related to the object’s own properties such as shape,texture orientation,and roughness,and is not greatly attenuated by the interference of scattered particles.In this paper,aiming at the problem that traditional optical imaging systems cannot effectively highlight the target based on light intensity information alone in a complex environment,combined with the latest research results of polarization imaging technology,a polarization imaging system based on ZYNQ is developed,and a polarization image fusion network model is designed to collect intensity images and polarization degree images by the fusion system,and the effectiveness of the fusion method is verified by experiments,and the related work can be summarized as follows:(1)The polarization state of light and the quantitative description method of polarized light are introduced,the advantages and disadvantages and application scenarios of four polarization imaging systems based on Stokes vector are analyzed,and combined with the research background of this topic,the split-focal plane polarization imaging system is selected to complete the project design.(2)According to the functional requirements of the system application scenario,the overall scheme of the hardware design of the polarization imaging system was formulated,the schematic diagram of each part of the functional circuit was designed using the idea of modularization,the PCB was designed in the form of a combination board,and the board making,welding and the construction of the entire system were completed.The length,width and height of the prototype are 66.40 mm,42.31 mm and 48.05 mm respectively,realizing the miniaturization design of the system.(3)On the basis of the system hardware design,the overall scheme of software design is formulated,and the software function is divided into five parts,such as CMOS initialization,image acquisition,image processing,image transmission and application initialization,and the functional design of each part is completed respectively,and the image acquisition function of the system is verified through polarization image acquisition experiments.The prototype has a resolution of 1920*1080 and a frame rate of 30 Hz.(4)Based on the characteristics of intensity image and polarization image and the rules of image feature extraction,a polarization fusion network is designed to integrate the intensity image and polarization degree image collected by the system.And the fusion effect of the network model is verified by subjective and objective evaluation compared with the existing polarization image fusion method.Compared with the improved dual-channel pulse-coupled neural network,which has the best fusion effect,the average gradient of objective evaluation index,information entropy,image quality,standard deviation and improved multi-scale structural similarity of the fusion method in this paper are increased by 4.3%,1.0%,8.1%,2.5%,3.1%,and image noise is reduced by 8.8%,respectively,the superiority of this method is verified.