A Study of the Performance of Motion Estimation in Vector Based Color Image Super-Resolution Analysis for Pure Translational Motion

Abstract

AbstractIn this paper, we have used the existing vector-magnitude based super-resolution algorithm and optical flow motion estimation method for super-resolution analysis of color images. In vector-magnitude based approach, all the three color bands are treated as vector instead of processing them separately. Color artifacts are reduced to a great extent when the pixels are treated as vectors and processed. The motion vectors are considered to be unknown and to estimate the motion, the Lucas-Kanade Optical flow method have been used. For over-determined systems, the Lucas-Kanade algorithm can be used in combination with statistical methods to improve the performance in presence of outliers as in noisy images. The advantage of this algorithm is the comparative robustness in presence of noise. The performance of the algorithm is analyzed and compared with the methods that uses known motion vector, by considering Mean Square Error (MSE) and Peak Signal to Noise Ratio (PSNR). The algorithm for deriving the super-resolution color images from a set of low resolution color images is implemented in the MATLAB and simulation results show the effectiveness of this method.KeywordsColor ArtifactsDiffraction LimitMotion EstimationOptical FlowOutliersOver Determined SystemsSuper-Resolution