Research On Battlefield Target Millimeter Wave Radiation Characteristic And Super-Resolution Algorithm Of Millimeter Wave Image

Passive millimeter wave imaging detection system by receiving scene outward radiation of electromagnetic wave,using radiation difference of different material for imaging.Because of its no radiation,non-contact,good concealment and can detect hidden items,so they are widely used in field detection,security checks,and other fields.However,passive millimeter-wave imaging system under the influence of antenna,channel characteristics,the Imaging results in fuzzy,distorted,and so on.This brings to the subsequent target detection and tracking processing the serious influence.So we need the image Super-resolution processing to improve picture quality.Based on the practical project,The major work of this dissertation include:1.The system of passive millimeter wave imaging detection system is studied,millimeter-wave Imaging degraded in the process model is analyzed.Thus established by millimeter wave radiation characteristic and image super resolution processing as the main research direction.2.Studied the propagation characteristics of millimeter-wave,focused on the clouds,and fog attenuation on millimeter wave propagation.And the relationship of metal material's emissivity and the incident angle is studied.Established a solid targets ' millimeter wave radiometric brightness temperatures and the simulation experiment was carried out.3.Studied the mathematical model of millimeter wave image super resolution algorithm based on preliminary classification and analyzed its main ideas.By using Gabor filter for feature extraction and classification using k-means algorithm,Shortcomings of traditional algorithm is improved.The results of simulation experiments show that the improved algorithm for optical and millimeter wave image has good effect of super resolution reconstruction.4.Studied the image super-resolution algorithm based on convolution neural network(SRCNN).Aimed at the high noise,low resolution and other characteristics of millimeter-wave image,I improved the original neural network structure.The simulation results show that the improved algorithm has achieved good results in the super-resolution reconstruction effect and real-time performance.