Images with higher resolution are required in almost all digital imaging applications.For past few decades considerable advancement has been realized in imaging devices.Modem digital cameras are equipped with high quality lenses, increased pixelresolution as well as small size, but still they are far from the perfection because ofbeing costly and having physical limitations of hardware for example sensor, lens andoptics. Using signal processing techniques to generate a high resolution image from alow resolution image or a set of low resolution images is a cheaper and effectivealtemative. Such kind of resolution enhancement is called super resolution imagereconstruction. The resolution enhancement is achieved by fractional-pixeldisplacements between observed low resolution images. Super resolution methodstend to overcome the limitations of an imaging system where there is no need foradditional hardware. This fact has made super resolution a hottest research area bothscientifically and commercially. The research work in this thesis is mainly focused onfour tasks.　　 Firstly, we present an overview of well-known super resolution techniques for bothmulti-image super resolution and single-image super resolution. The performancecomparison and analysis are the main concerns to see the advantages anddisadvantages of multi-image super resolution techniques. After applying the superresolution techniques to our tested synthetic image data we critic the affect of the fullreference quality metrics (peak signal-to-noise ratio and mean square error).　　 Secondly, an optimized approach is introduced for the effective selection of lowresolution images in the process of super resolution image reconstruction. We used mimmum two low resolution images and performed experiments to validate ourmethodology.　　 Thirdly, we investigate the non-parametric super resolutions for object recognition inRadar. We analyze how super resolution algorithm can be useful for resolving two or more targets in radar imaging system. The performed simulation results help us toevaluate the performance of non-parametric super resolution algorithms discussed.Finally, a little research has been carried out about a proposal for futuristic design ofimplementing super resolution image reconstruction in the environment of mobilecloud computing. The proposed framework is based on the idea of utilizing the cloudcomputing infrastructure through the mobile device + mobile web (or intemet serviceprovider) as an Educational Tool for super resolution image reconstruction.Throughout this research work, experiments on vanous real and synthetic image dataare conaucted to validate and evaluate the performance of the existing and proposedappro aches.