Generation of super-resolution video from low resolution video sequences: A novel approach

Abstract

The term "super-resolution" refers to the process of obtaining higher-resolution images from several lower-resolution ones, i.e. resolution enhancement. The quality improvement is caused by sub-pixel shifted from each other between images. Each low resolution image contains new information about the scene and super-resolution aims at combining these to give a higher resolution image. Super-resolution allows us in overcoming the limitations of the imaging system (resolving limit of the sensors) without the need for additional hardware. This paper deals with high resolution video reconstruction from low resolution video. An algorithm for enhancing the resolution of video through histogram based segmentation and frequency domain registration is proposed. Segmentation of the video is done using Histogram method and frequency domain approach is used for registration of images. Bi-cubic interpolation is applied to this video to generate the super-resolution (SR) video. Based on the requirements additional brightness and temporal resolution can be added to the SR video. It is tested on indoor/outdoor real video data, demonstrating the feasibility of the approach, and then results are presented. Qualitative analysis is presented to analyze the quality of reconstructed Super resolution video comparing to input Low Resolution video. The results of this work can be used for object tracking and identification. � 2007 IEEE.