Antipodally Invariant Metrics for Fast Regression-Based Super-Resolution

PROJECT TITLE :

Antipodally Invariant Metrics for Fast Regression-Based Super-Resolution

ABSTRACT:

Dictionary-based super-resolution (SR) algorithms sometimes select dictionary atoms based on the gap or similarity metrics. Though the optimal selection of the nearest neighbors is of central importance for such methods, the impact of using proper metrics for SR has been overlooked in literature, mainly due to the vast usage of Euclidean distance. In this paper, we tend to present a terribly quick regression-primarily based algorithm, that builds on the densely populated anchored neighborhoods and sublinear search structures. We have a tendency to perform a study of the nature of the features commonly used for SR, observing that those features sometimes lie in the unitary hypersphere, where every point features a diametrically opposite one, i.e., its antipode, with same module and angle, however the other direction. Even though, we validate the benefits of using antipodally invariant metrics, most of the binary splits use Euclidean distance, that does not handle antipodes optimally. So as to learn from each the worlds, we propose a easy yet effective antipodally invariant remodel that may be simply included within the Euclidean distance calculation. We tend to modify the first spherical hashing algorithm with this metric in our antipodally invariant spherical hashing scheme, obtaining the identical performance as a pure antipodally invariant metric. We have a tendency to spherical up our contributions with a novel feature remodel that obtains a higher coarse approximation of the input image because of iterative backprojection. The performance of our methodology, which we have a tendency to named antipodally invariant SR, improves quality (Peak Signal to Noise Ratio) and it's faster than any different state-of-the-art method.

Did you like this research project?

To get this research project Guidelines, Training and Code... Click Here