Abstract
This paper is a follow-on to a previous study that comprehensively tested higher order polynomial, calibration-based, non-uniformity correction (NUC) algorithms targeted at short wave infrared (SWIR) imagers. As part of the previous study, a new nonlinear gain equalization algorithm was developed and compared with existing algorithms. Infrared imaging is an expansive field with many commercial and military applications. A known problem with infrared imaging devices is their non-uniformity. Contributing factors of non-uniformity include, but are not limited to, the unique photo response of each detector, amplifier mismatch, and dark current. This study takes a higher level view with the objective of determining the impacts of preprocessing on performance of the recently developed gain equalization algorithm. A basic linear banding threshold using mean replacement is compared with an unsupervised machine learning algorithm using principal component analysis (PCA) and nearest neighbor median filtering for replacement The details of both approaches are explained in this paper. Standard performance metrics such as the root mean square error (RMSE) and the residual non-uniformity (RNU) are compared. The techniques used in this study may be applied to any wavelength device, however testing was done using multiple sets of raw focal plane array data collected from SWIR cameras. Similar experiments could be performed on simulated data sets for reproducibility purposes and details on the experimental setup are included for this purpose. Experiments were repeated for different gain settings and device resolutions and the results were compared with other calibration-based nonlinear NUC algorithms. By applying preprocessing techniques, such as bad pixel identification and replacement, the accuracy of image data is improved and RMSE reduced by additional 5% on average. For certain applications even small levels of improvement are significant and finding algorithms that are easy to implement can be essential. This study highlights the improvements that can be made to SWIR devices by applying simple preprocessing techniques. As the use of SWIR technologies increases, performance demands are also expected to increase, thus making studies such as this more and more significant.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.