Vector-magnitude based super resolution for color images

In the recent years charge coupled devices (CCDs) and CMOS image sensors are widely being used for capturing a digital image. The current resolution level will not satisfy the requirements for certain applications. One of the solutions for satisfying all requirements is Super Resolution. Super resolution is a process of increasing the density of pixels by extrapolating the spectrum of an object beyond the diffraction limit of the imaging system. In this paper, we address the problem of colour image super-resolution of the existing scalar spatial domain methods, Least Square method and Robust Super Resolution Method, where all the three colour bands are separately treated. So in the Vector-Magnitude Based Super Resolution algorithm we treated all the three colour bands as vector instead of processing all the three colour bands separately. Colour artifacts can be reduced, when colour bands are treated as vectors and processed. That is, it is enough to apply super-resolution algorithm to the luminance component and a suitable interpolation method to the chrominance components of the RGB colour image. For this, the representation of the colour image in terms of the magnitude and the orientation is first considered. Then the magnitude part is super-resolved and the orientation part of the super-resolved point is replaced by orientation of bilinear interpolated orientation of any one of the low resolution image. And we analysed this algorithm performance with the scalar methods by considering execution time, MSE (Mean Square error) and PSNR (Peak Signal to Noise Ratio).In this paper, super resolution of colour images is implemented in the MATLAB.