The Study Of Infrared Focal Planearray Non-uniformity Correction Algorithm Based On Optimized Kalman Filtering

Infrared Focal Plane Arrays(IRFPA)with advantages of high sensitivity, highratio of signal to noise, simple and compact in structure, is the most future developinginfrared imaging device. However, due to the restriction of semiconductor materialand technology, there are some difficulties to make IRFPA has uniform response. Sobased on the modern signal processing technology, we can solve the non-uniformityresponse problem by algorithm, which has low cost, high efficiency, important valuein application and engineering. After decades of deeply researches and explorations,there have been proposed many non-uniformity correction(NUC)algorithms. But eachcorrection algorithm is subject to the certain environment or occasion, hard to meetthe requirement for common application.Firstly,this paper analyzes the research value and current situation of NUCalgorithms for IRFPA, and the main problems of these algorithms. Then, thecharacteristics of IRFPA and generation mechanism of non-uniformity are studied,and the shortages of the NUC algorithms are verified by simulating for application.Thereby, this article studies a new NUC algorithm of IRFPA based on Kalmanfiltering, which is the focus in the scene correction algorithm with advantages of anaccurate tracking and estimating ability on response parametric variation. However,the classic Kalman filtering non-uniformity correction algorithm has a small dynamicrange, low accuracy of correction and larger calculation. Aiming to solve theseproblems, the classic algorithm has been improved and optimized,and then the paperproposes an optimized non-uniformity correction algorithm. It simplifies the classicKalman algorithm, reduces the calculation and elevates the operating efficiency byusing the inverse matrix of error covariance to do the iteration.The end of this paper shows the simulatiing experiments of all relatedalgorithms with Matlab. The experimental results validates the conclusion that theproposed NUC algorithm can not only achieve linear and non-liner non-uniformity onthe focal plane array detector, but also improve correctting speed effectively. Inparticular, when the number of correctting frame increases to1000, the correcttingspeed can be enhance nearly2-fold, which proves the superiority and feasibility of theoptimization algorithm.