PROJECT TITLE :

Dissecting and Reassembling Color Correction Algorithms for Image Stitching - 2018

ABSTRACT:

This Project introduces a new compositional framework for classifying color correction strategies per their 2 main computational units. The framework was used to dissect fifteen among the best color correction algorithms and also the computational units so derived, with the addition of four new units specifically designed for this work, were then reassembled in a very combinatorial way to originate concerning 100 distinct color correction methods, most of which never thought-about before. The higher than color correction ways were tested on three completely different existing datasets, including each real and artificial color transformations, plus a completely unique dataset of real image pairs categorized in keeping with the kind of color alterations induced by specific acquisition setups. Differently from previous evaluations, special emphasis was given to effectiveness in globe applications, like image mosaicing and stitching, where robustness with respect to strong image misalignments and lightweight scattering effects is needed. Experimental evidence is provided for the first time in terms of the most recent perceptual image quality metrics, that are known to be the closest to human judgment. Comparative results show that mixtures of the new computational units are the foremost effective for real stitching situations, irrespective of the precise supply of color alteration. On the other hand, within the case of correct image alignment and artificial color alterations, the best performing methods either use one amongst the new computational units, or are created of contemporary combos of existing units.