Research On Comprehensive Parameter Testing Technology Of Mid- And Far-infrared Hyperspectral Loading

The mid-and far-infrared imaging system has important application value in many fields due to its characteristics of covering a long band,strong penetrability,and the ability to detect temperature information of objects.Before the mid-far infrared hyperspectral load is lifted off,or before any infrared camera product is used,it is necessary to evaluate its parameter performance indicators.However,the infrared imaging system is faced with many parameters,complex and expensive test systems,and imperfect test methods.In order to solve these problems,this paper studies the testing technology of the comprehensive parameters of the midfar infrared hyperspectral load.This paper divides the comprehensive parameters into spectral parameters(relative spectral response rate)and imaging parameters(modulation transfer function(MTF),noise equivalent temperature difference(NETD),minimum resolvable temperature difference(MRTD),image distortion and field of view).Based on the relative spectral response rate calibration link of the diffuse reflection integrating sphere,the evaluation models of MTF,NETD and MRTD parameters were constructed,the MTF test method based on the edge diffusion function was optimized,and the NETD parameter test based on the signal transfer function was implemented.Based on theoretical research,the principle prototype is verified.The test level of MRTD and NETD parameters is equivalent to the imported CI-Systems test system,which verifies the stability and practicability of the principle prototype.Based on the principle prototype,a comprehensive parameter test system for infrared loads in a space environment is designed and built to simulate the space environment.A point source black body and a monochromator combined with an integrating sphere are used to generate high-precision mid-far infrared monochromatic light.The off-axis reflection type parallel light pipe was used to simulate infinite targets,and the system control and data processing software based on Labview was developed.Finally,the comprehensive calibration of spectral parameters and imaging parameters was completed.Aiming at the limitations of traditional MRTD test methods in infrared imaging systems,such as long test time,high uncertainty in test results,and high labor costs,this paper proposes a MRTD parameter test method based on CNN neural networks.The Four Le Net network model was proposed by streamlining the network structure and optimizing the activation function.It obtained a 94.1% correct recognition rate in the training of the enhanced four-shot target image data set,which is higher than other current network models.This method shortens the test time on the basis of guaranteeing subjective judgment,and effectively reduces labor costs.In order to ensure the accuracy of the test system and the integrity of the test results,this paper uses the GUM method to analyze the uncertainty of the test results of the comprehensive parameters,and clarifies the source of the uncertainty.The standard synthetic uncertainty of the relative spectral response rate is related to the wavelength,the standard synthetic uncertainty of MTF is 1.8%(k =?3),and the standard synthetic uncertainty of NETD is 0.87%(k =?3).Finally,this paper analyzes the impact of image noise processing and image enhancement processing on the test results of imaging parameters.The analysis results show that the median filter and Gaussian filter have an optimization effect on MTF.In the NETD test process,it is necessary to turn off the automatic gain and adjust the gamma correction to the linear state to improve the test accuracy.