Research On Resistor Array Non-uniformity Correction With Adaptive Correction Algorithm

Resistor array as the core device of the Infrared Imaging Simulation System, itsnon-uniformity directly impacts imaging results of the system. Over the years,researchers both at home and abroad have given high priority to improve thenon-uniformity of the array and actively carrying out research. But non-uniforminformation collection accuracy is still low, the calibration work is complex,hardware implementation is difficulty. These deficiencies affect the uniformity ofthe array radiation imaging. This has great significance to the applications ofInfrared Imaging Simulation System.In this paper, further study in non-uniform data measurement andnon-uniformity correction algorithm had been done. Studied different correctionalgorithms to improve the non-uniformity situation and find out its applications onthe basis of accurate non-uniform information. The main work includes thefollowing sections:At first, did research in obtaining accurate original non-uniform informationmeasurement method, compared simple low-cost sparse grid method, the simplifiedsparse grid method, efficient PSF roughly estimation full screen litting method andsparse grid-PSF estimation hybrid test method. Study different test methods toachieve the accuracy requirement and find out the applicable test environment. Gotaccurate original non-uniform information for non-uniformity correction inpreparation.Besides, proposed segmentation correction algorithm, gave out the resistanceelement characteristic response curve fitting method. Complete characteristic curvelinearization and other segment correction basic work. Then, studied segmentationcorrection algorithm principle and realization process, and simulated the algorithmwith value resistor. Draw out that segmented correction algorithm had a very goodimprovement for non-uniformity.And then, proposed adaptive correction algorithm, set up algorithm networkmodel structure, determined the activation function, and the objective function, andthe weights and bias modification method. And use the model to train out the outputresponse curve. Gave the principle and implementation of the adaptive algorithm.and simulated the algorithm with value resistor. And obtained in the ideal case ofinherent non-uniformty, the adaptive correction algorithm has a very goodcalibration results.At last, Build FPGA-based hardware verification platform. Embedded thesegment correction algorithm and adaptive correction algorithm in the verification platform, to measure the correction accuracy of the algorithm for the inherentnon-uniform nature with the specific indicators and the observed actual correctionresults. And accuracy of different correction algorithms can be achieved, aprerequisite of the application and the possible applications.