Empirical filter estimation for subpixel interpolation and matching

Abstract

We study the low-level problem of predicting pixel intensities after subpixel image translations. This is a basic subroutine for image warping and super-resolution, and it has a critical influence on the accuracy of subpixel matching by image correlation. Rather than using traditional frequency-space filtering theory or ad hoc interpolators such as splines, we take an empirical approach, finding optimal subpixel interpolation filters by direct numerical optimization over a large set of training examples. The training set is generated by subsampling larger images at different translations, using subsamplers that mimic the spatial response functions of real pixels. We argue that this gives realistic results, and design filters of various different parametric forms under traditional and robust prediction error metrics. We systematically study the performance of the resulting filters, paying particular attention to the influence of the underlying image sampling regime and the effects of aliasing ("jaggies"). We summarize the results and give practical advice for obtaining subpixel accuracy.

Spatially-Varying Image Warping: Evaluations and VLSI Implementations

“…Our observations are in line with claims and evaluations from previous works [1,10,18,20]. Quality-wise, we conclude that the adaptive EWA splatting approach is superior both to simple bilinear interpolation and bilinear interpolation with super-sampling.…”

Rapid, detail-preserving image downscaling

“…Therefore, different approaches for this scenario have been investigated. Triggs [2001] proposed an approach to design filters that allow for reconstruction of the high resolution input image with as little error as possible. This, however, leads to the same type of filters as frequency-based approaches so that small details of the original image are still lost.…”

Image template matching using Mutual Information and NP-Windows

“…Better approximations may be obtained by prior knowledge of the sampling pre-filter or using empirical estimation [10]. For nearest neighbour interpolation this method breaks down to standard sampling.…”