Research On Imaging Performance And Image Processing Technology Of Complex Optical Machine Functional Parts

Complicated optical-mechanical image information collection technology has made great progress due to the continuous deepening of aviation technology.In the process of collecting information,the optical system of the carrier is affected by the size of the field of view,temperature and stray light,and there are often interference from factors such as camera vibration and clouds during the image acquisition process.The quality of the recorded data set limits its subsequent analysis and application.Therefore,there is an urgent need for an effective imaging system to acquire the data.At the same time,effective processing methods are required to improve the quality of the data set.The main content of this disserta tion is divided into two parts: taking the self-developed optical imaging system as the research object,analyzing the performance of the system and observing its sample quality;the other part is to conduct in-depth research on the noise pollution,the lack of important information and the lack of prominent key feature information in remote sensing images.(1)The research object is an optical system developed in-house,and the system is analyzed to theoretically check whether it meets the requirements for retrieving information from a complex optical machine.An experimental device is set up to obtain image data under various conditions and relevant professional software is used to analyze the detailed information of the sample.Compare the analysis results,evaluate the system's performance,and observe the quality of its evidence.(2)Traditional wavelet transform(WAVELET)targets poor image quality obtained by a complex optical and mechanical optical imaging system,and cannot effectively express image boundaries,transition zone,and other information.The conventional noise elimination function cannot discriminate effectively between noise,and new remote sensing information is proposed to eliminate other defects.In noise reduction processing,the optimized contour transform is used to replace the traditional wavelet,the threshold function is optimized,the validity of the function is checked,and the sample obtained from the self-developed optical imaging system is used and the sample is used as an experimental one.(3)A new method of remotely sensing image reconstruction is proposed to solve the problems of information loss due to noise and dead pixels during the process of obtaining evidence by an optical imaging system of an imaging system,which reduces image quality and limits accuracy.Ring tensor theory is the basis of the scheme and Bayesian probability is fed into the ring tensor decomposition model to predict missing information in the proofs.In the data space,the adaptive regularization of the kernel norm is included in the processing of low rank factors,and the decomposition of singular values is performed on a more precise scale,which effectively reduces the burden of selection of rank.Finally,simulated data and samples obtained with a sophisticated optomechanical imaging system are used as experimental data to validate the proposed scheme.(4)A method is proposed for solving the problem that traditional methods of image improvement cannot effectively integrate information about samples,and often have low image quality and insufficient important characteristic information,which significantly limits the accuracy and efficiency of extracting useful information in the image.The method takes the concept of a tensor as a whole,designates the samples to be processed to varying degrees,injects and unfolds several previous data,collects a series of images after noise reduction,builds a high-order tensor model,and conveys the Bayesian probability.The ring tensor model predicts and augments information such as important features,detailed textures,and contours.Finally,experiments are carried out to demonstrate the circuit.