Image Capture Modelling for High Resolution Reconstruction.pdfVIP

38 Image Capture Modelling for High Resolution Reconstruction Donald G Bailey Institute of Information Sciences and Technology, Massey University, Private Bag 11222, Palmerston North, New Zealand Abstract: A research topic that has been receiving much interest recently is the reconstruction of a high resolution image from an ensemble of low resolution images of the same scene. Such image fusion implies that each low resolution image individually does not contain all of the available information, and therefore must be subject to aliasing. The reconstruction process involves resampling the ensemble at a higher sample rate, effectively unscrambling the aliased information. However, the reconstructed high resolution image appears blurred because of area sampling caused by the sensor, combined with the low pass response of the camera electronics. By modelling the image capture system, the point spread function may be estimated and then removed by using inverse filtering in the frequency domain. If the individual low resolution images are subject to a spatially variant point spread function (resulting from perspective or radial distortion or focus limitations), a more complex spatially variant inverse filter must be used. Keywords: image fusion, super-resolution, aliasing, inverse filtering, camera calibration 1 Introduction Image fusion involves the combining of data from a number of image sources in such a manner that the output image contains more information than that contained in any of the individual images. The image sources are often different in their properties or characteristics. An importa