Research On Performance Evaluation Model And Image Processing Method For Terahertz Imaging System

Terahertz has many unique features such as transparency,broadband,high temporal and spatial coherence,low energy and unique transmission characteristics,it shows a wide range of applications in safety detection,spectroscopy,imaging and communications,chemistry,biology,materials science and pharmacy and other fields.As the main application mode of terahertz technology,terahertz imaging has become the key technology and frontier research direction of China and international.However,it is limited by the development of technology.At present,the research on terahertz imaging technology is mainly focused on the radiation source,and the detector.The main problems of the terahertz imaging technology are: the lack of effective imaging theory to guide the design of the imaging system;the unevenness of the radiation source and the limitation of the material of the detector makes the terahertz focal plane image have serious nonuniformity noise;terahertz and millimeter-wave dual-band passive scanning imaging can compensate for terahertz radiation weak atmospheric permeability,for millimeter-wave low resolution,and to obtain the fusion image containing more scene information,but because of the restrictions of the detector and the optical system,there is a large amount of noise and affine transformation in the dual-band image of terahertz and millimeter-wave,and the fusion image can not be obtained directly by using the fusion algorithm.To solve these problems,we need to make breakthroughs and innovations in a number of technologies,such as system performance evaluation model and terahertz image processing algorithms such as non-uniformity correction,image preprocessing and fusion,which are the current frontier direction of terahertz imaging research both at home and abroad.In this paper,the terahertz imaging theory and image processing algorithms for solving the above problems are studied in detail.In particular,the main contents and innovations are summarized as follows:(1)Based on the characteristics of the vanadium oxide-based uncooled terahertz focal plane detector,an uncooled terahertz focal plane detector assembly was designed.In view of the lack of the theoretical guidance in the design of the imaging system and optical systems,the system design theory of terahertz active imaging system based on uncooled focal plane detector is studied.In the limit of the terahertz focal plane detector noise equivalent power and terahertz radiation source power,the target and scene characteristics,optical components and atmospheric attenuation and other factors are analyzed synthetically,imaging area and image contrast model was proposed for characterization of terahertz focal plane active imaging system.According to the specific example of the terahertz focal plane active imaging system,the verification results are in accordance with the experimental results of the reference system,and the validity of the model is verified.It is proved that under the existing conditions,the effective imaging area can only meet the detection of the size of the envelope Only to increase the energy of the terahertz radiation source or further use of the machine scanning system,the body body / body,suitcase and other tests can be met.The results provide a sufficient theoretical basis and technical support for the design of the terahertz focal plane active imaging system.(2)In the terahertz active imaging process,not only the detector noise equivalent power,radiation source power,optical element attenuation,atmospheric attenuation and so on will affect the imaging results,the transmission characteristics of detectors,optical components,displays and human visual will also affect the system performance.Aiming at the problem of performance evaluation of terahertz active imaging system,based on the analysis of the external factors of the system,combined with the transmission characteristics of the components,especially the fixed pattern noise caused by uneven and strong coherence of the terahertz radiation,the noise sub-models are improved,and the signal distribution and system transfer characteristics in the system are analyzed,the minimum resolovable contrast matched filter model is proposed.Aiming at the model,the specific system design examples and analysis are carried out,and the validity of the model is validated in theory and experiment.This model can provide a more effective theoretical basis and performance evaluation parameters for the design of the terahertz active imaging system.(3)This paper analyzes the nonuniform noise sources and manifestations of terahertz focal plane images,and analyzes the edge loss,smear and ghosts in the nonuniformity correction of the terahertz focal plane image by the traditional scene-based nonuniformity correction algorithm.The neural network nonuniformity correction algorithm based on improved guided filtering is studied in order to better protect the edge and detail of the image and improve the smear and ghosting while preserving the nonuniformity of the image.On the basis of the above research,aiming at the noise of the terahertz focal plane image,the denoising process including blind element compensation,time domain filtering and nonuniformity correction is designed to process the terahertz focal plane imaging to obtain a better quality image.(4)At present,scan-based terahertz imaging technology based on unit / multi-detector has been gradually applied.In order to improve the efficiency of target detection,dual-band and multi-band terahertz imaging technology is developing rapidly,in which terahertz and millimeter-wave dual-band imaging can compensate for the shortcomings of single-band imaging and obtain a fusion image with more scene information relative to a single information image.However,due to the low power of terahertz and millimeter wave radiation in nature,the signal-to-noise ratio of terahertz and millimeter-wave images is low,and the spatial resolution of imaging system is also low due to the limitation of scanning mode,resulting in the low clarity and contrast of the dual-band terahertz and millimeter-wave image.Image denoising,enhancement and other image preprocessing algorithms,as well as image registration and fusion to show multi-band imaging advantages of image processing methods are urgently needed.This paper comprehensively analyzes the imaging modes and image noise characteristics of the existing terahertz and millimeter-wave dual-band passive security imaging system.A terahertz and millimeter-wave dual-band image preprocessing and fusion algorithm are proposed,which combines the three-dimensional block matching filtering,histogram equalization,image registration based on mutual information,and image fusion based on wavelet transform.The algorithm can effectively reduce the noise of the human security images,improve the contrast of image detail,and realize the fusion image,reduce the false detection rate and false alarm rate.In addition,the minimum detectable temperature difference test method is used to evaluate the performance of the system objectively,and it provides the key technologies and tools for the practical evaluation and performance evaluation of the dual band imaging system.In this paper,the research on the key scientific problems such as imaging theory and image processing of terahertz images has made obvious progress.For the development of terahertz imaging device with independent intellectual property rights,imaging method and theory,core technology,and enhance the technical level and innovation of terahertz imaging in China,and the ability to occupy a place in the field of international terahertz core technology has important significance and role.