Edge highlighting of laser-scanned point clouds improves the accuracy of perceived depth in transparent multi-view 3D visualizations

Abstract

The rapid development of 3D scanning has allowed us to digitally capture cultural heritage objects as large-scale point clouds, recording their complex internal and external structures. We have developed a 3D multi-view transparent visualization using edge highlighting as visual guide, improving accuracy of perceived depth for reliable visual analyses of the 3D scanned point clouds. When transparently visualizing a 3D object on a 3D multi-view display, the depth information tends to become less distinct. Therefore, we aimed to explore whether emphasizing the statistically extracted 3D edges of the scanned object can enhance the perceived depth of transparently visualized 3D objects. We conducted psychophysical experiments where participants were asked to report their perceived magnitudes of depth for multiple types of 3D structures of cultural heritage presented on a multi-view 3D display. The results indicated that our edge highlighting reduced the underestimation of depth from 46.8% to 34.0% on average. This improvement persisted even when the 3D structures produced an atypical linear perspective, normally caused by reverse depths, where the front and back are reversed. These findings suggest that edge highlighting brings us an effective visual guide to improve the accuracy of perceived depths on a 3D multi-view display.