Abstarct: Super-resolution (SR) is a promising technique of digital imaging which attempts to generate a raster image with a high resolution from low resolution input image(s). This thesis focuses on multi-frxames super resolution. In interpolation methods increasing resolution is performed by up sampling the image. These methods only increase number of pixels of image and missing details are not reconstructed. Super-resolution methods are proposed for solving the problem of interpolation methods. Super-resolution technique not only improves the quality of under sampled, low resolution images by increasing their spatial resolution but also attempts to filter out distortion. Classic SR methods do not produce a good result on actual videos, because these methods require accurate motion estimation. New methods are introduced in video super resolution that can produce effective results without accurate motion estimation, recently. SR method baxsed on NLM filter use fuzzy motion instead of exact motion estimation. In this thesis we propose a new method baxsed on probability. In proposed method, by introducing the concept of probabilistic registration, fuzzy motion that has been introduced in earlier work is improve. In the proposed method using other video frxames we can restore the missing high resolution details. baxsed on proposed method user can overcome problems in imaging systems using moderate cost hardware. Experimental results show that the proposed method can remove artificial effects and produce sharp edges. The proposed method in terms of PSNR criteria is better than the other methods introduced in this thesis.