Smart onboard image enhancement algorithms for SWIR day and night vision camera

摘要

SWIR imaging based on InGaAs based FPAs is well suited for passive or active day and night vision applications in different weather conditions, including surveillance, defense or fire-fighting. Xenics developed the Rufus camera, based on a 640 × 512 pixel resolution FPA. In order to achieve the best performance over a large span of lighting conditions, different smart algorithms are implemented onboard. The auto-exposure algorithm optimizes the integration time in order to position the image histogram at a given user-controlled brightness level. Moreover the algorithm can also switch automatically between different gain and read-out modes. At the same time a TrueNUC™ algorithm is calculating the non-uniformity correction. This correction depends on the detector temperature and integration time, because of the variable dark current of the InGaAs diodes. After the image correction and auto-exposure, further image enhancement is done by additional auto-gain and histogram equalization algorithms. Depending on the application, the user can modify several parameters of the algorithms, e.g. the maximal allowed stretching, the output histogram position and equalization strength. In the paper we will report on the performance of the algorithms at different environmental conditions. The residual Fixed Pattern Noise (FPN) of the TrueNUC™ model is analyzed. For the TrueNUC™ implementation a typical residual FPN of ＜1% is obtained (at 25℃) over the complete integration time range from 100us up to 40ms, both in high and low gain. Finally we will illustrate the capabilities of the algorithms in different applications.