Comparison of Normalized Transfer Functions for Fast Blending-based Color Correction of Scans Acquired Under Natural Conditions

Abstract

This work deals with the multiview colorization of large site acquired during rapid survey conducted in natural uncontrolled climatic and light conditions. Image stitching and weighting methods are adapted to perform fine per vertex depth maps colorizations. Different transfer functions are evaluated to obtain the best visual results. The rendering computation time of our solution is compatible with real-time fieldwork. For models composed of large scans series, the results exhibit an uniform coloring and no significant visual artifacts. A global color quality map is also produced to identify weakness and to determine the next acquisition parameters.