A Robust Multi frame image Super Resolution Reconstruction (SRR) is a process which produces a better or superior quality, High Resolution (HR) image from multiple numbers of blurred noisy low resolution (LR) images of the similar scene, acquired under different conditions. It produces a high quality super resolution image with higher spatial frequency, reduced noise and image blur as compared to the original images. The inputs images can be in the form of medical images, surveillance footage, digital video, satellite terrain imagery, or images from many other sources. In many visual applications like military and civilian areas the imaging sensors used are having very poor resolution outputs. Due to some limitations i.e. cost or hardware, resolution cannot be improved by replacing imaging sensors, hence super resolution algorithms plays a crucial role in improving the resolution of the output image. The Super resolution (SR) algorithms can be implemented in such cases as SR algorithms is a low cost algorithm and easy to implement. Super Resolution Reconstruction (SRR) is a computationally intensive process. The purpose of this paper is to develop a super resolution reconstruction algorithm which can be useful to real time data and data generated synthetically. A fast and robust technique of Multi frame Super Resolution Reconstruction is presented where frame to frame motion estimation and image registration is developed iteratively using Lucas Canada Pyramidal optical flow algorithm. For the iterative approach, the high resolution image estimate obtained is the solution of Median Shift and Add method, along with Gaussian filter and down sampling. Further iteratively gradient back projection and gradient regularization is carried out to obtain High Resolution (HR) image. A MATLAB based GUI for Super Resolution Reconstructed image is developed which calculates the image quality parameters such as PSNR, RMSE, SD, SSIM and percentage change in pixels value in HR image with respect to original frame. A PSNR plot for different test vectors is plotted with respect to number of frames and number of iterations. A MATLAB based tool is also developed for synthetic data generation. The HR image of different test vectors (real and synthetic) are shown in the form of MATLAB based GUI developed with and without adaptive filters. The change in the image quality parameters like SD and SSIM with and without adaptive filters is also shown in the form of table. The proposed method and its experimental results show that there is significant improvement in image quality, maintaining the edge information and has a higher PSNR and better visual effect. © 2015 IEEE.