Methods and algorithms for image fusion and super resolution Master of Science Thesis

This project is focused on image fusion of images of di erent focus depth and fusing for super resolution. The aim is to study these concepts and provide simulations and evaluations on various implementations. When performing multi focal fusion the images are decomposed by wavelets to obtain high frequency coe cients which is used to determine which parts of the input images that makes it into the fused image. The same technique is tested on images of di erent modality. Super resolution utilizes measurements of subpixel shifts between several low resolution images of the same scene to create a fused image of higher resolution by interpolation and image enhancement. The thesis describes a modular approach for super resolution where registration, interpolation and blind deconvolution is treated as separate modules. Tests are performed for di erent images, choices of modules and input parameters. List of abbreviations CT Computerized tomography HR High Resolution LR Low Resolution MBD Multiframe Blind Deconvolution MR/MRI Magnetic Resonance (Imaging) PSNR Peak Signal to Noise Ratio, evaluation SNR Signal to Noise Ratio SR Super Resolution