Optimized NCC-information theoretic metric for noisy wavelength band specific similarity measures

Image registration, in general, offers an increased field of view from an airborne sensor which provides an advantage in intelligence, surveillance and reconnaissance (ISR) applications for border security scenarios. Normalized Cross Correlation (NCC) is one of the most precise area based template matching technique employed for image registration. In an operational sensor the image acquired will generally contain both system and external noise present in the field of view of the sensor. This, in turn, implies that the NCC may not provide the optimum match for registration. In this paper we present a technique that identifies an optimized correlation coefficient value or a set of coefficient values in the multi-spectral and hyper-spectral cases. In this paper NCC is used in conjunction with information theoretic measures to determine an optimized match, when band varying noise is also present. In particular, the proposed approach is based on a hybrid and combined NCC and information theoretic measures, and does not necessarily use a threshold value to identify potential matches based on the similarity measures. This work also analyzes the effects of varying the template size and the template scene dynamics on template matching accuracy. Finally, the technique is applied to hyper-spectral images, and the band sensitive response for the NCC performance is also determined.