Image registration for multi focus and multi modal images using windowed PCA

Due to the limited depth of focus or long focal lengths, it is not possible to get an image which contains all relevant objects in focus. The solution for this problem is to acquire several images with different focus points and registered them. windowed based principle component analysis is implemented for multi focus and multimodal images. Experiments are carried out on medical images of brain angiography and images for course of carotid arteries from ascending aorta to brain, required for looking for blockage in case of brain stroke, Multi modal medical images, taken from different sensors like CT and MRI images are registered using this algorithm for medical diagnostics. Multi modal navigation aid images for helicopter pilots taken with low light television sensor (LLTV) and thermal imaging forward-looking-infrared (FLIR) sensor are also registered. This is used for helicopter pilots for navigational aids. The proposed algorithm of windowed PCA is compared with conventional averaging method and PCA based method without windowing using different quality measures like Entropy, Correlation Coefficient, Histogram Error, Root Mean Square Error, Maximum Absolute Error (MAE), Peak Signal to Noise Ratio (PSNR), Standard Deviation and Universal Image Quality Index (UIQI).