Visual Scan-Path based Data-Augmentation for CNN-based 360-degree Image Quality Assessment

360-degree Image quality assessment (IQA) is facing the major challenge of lack of ground-truth databases. This problem is accentuated for deep learning based approaches where the performances are as good as the available data. In this context, only two databases are used to train and validate deep learning-based IQA models. To compensate this lack, a dataaugmentation technique is investigated in this paper. We use visual scan-path to increase the learning examples from existing training data. Multiple scan-paths are predicted to account for the diversity of human observers. These scan-paths are then used to select viewports from the spherical representation. The results of the data-augmentation training scheme showed an improvement over not using it. We also try to answer the question of using the MOS obtained for the 360-degree image as the quality anchor for the whole set of extracted viewports in comparison to 2D blind quality metrics. The comparison showed the superiority of using the MOS when adopting a patch-based learning.

Accuracy of Different Head Movements of Intraoral Scanner in Full Arch of Both Maxilla and Mandible

This in vitro study aimed to compare the accuracy of maxilla and mandible full-arch scans from an intraoral scanner via one scan path with six different head movements. Standard maxilla and mandible models via holder were set in a dental chair to simulate position and posture. The reference models’ standard tessellation language (STL) files were formatted via desktop scanner, and operative models’ files were obtained via IOS TRIOS 3 Pod as superimposed by Exocad CAD software. The same scan path with six head movements (Linear, Circle, Wave, 8-figure, S-figure, and A-P) were designed to scan 10 times per jaw, and a total of 120 scan files were then compared with reference files. The data were recorded and deviations of both occlusal and B-L sides were compared, with statistical analysis being performed by one-way analysis of variance (ANOVA) and post hoc comparisons with Tukey test. The trueness of optical impression for full-arch via one path with six head movements differed from maxilla and mandible and depended on the different movements (p < 0.05). In the same path, peak deviation was found at the turning points for left central incisor, left first premolar, left second molar, and right second molar in maxilla, and for right first premolar, second molar, and left second molar in the mandible. The 8-figure movement showed the highest deviation (0.128 ± 0.086 mm) in the maxilla and (0.105 ± 0.069 mm) in the mandible. The Linear movement presented the lowest deviation (0.096 ± 0.07 mm) in the maxilla while the Circle movement presented the lowest deviation (0.073 ± 0.041 mm) in the mandible, with the 8-figure movement showing the worst precision among six movements. In the maxilla, the S- and 8-figure movements were not recommended, while the Linear and Circle movements showed high trueness. In the mandible, the 8-figure movement was not recommended, while the other five presented similar lower deviations.

Visual Performance Assessment of Videos—A Case Study of the Game “Spot the Difference”

“Spot the Difference” is a well-known game where players must find subtle differences between two almost identical pictures. If “Spot the Difference” is designed for videos, what is the difference between videos and pictures? If the performance of videos is measured by an eye tracker, what scan paths will be conducted? In this study, we explored this game using a video to conduct a visual performance evaluation. Twenty-five subjects were recruited in a full-factorial experiment to investigate the effect of background (with background, without background), video type (animation, text), and arrangement (left-to-right, top-to-bottom) on searching, eye tracking performance, and visual fatigue. The results showed that the video type had a significant effect on the accuracy and subjective visual fatigue, with the accuracy and subjective visual fatigue for animation being better than for text. The results also indicated that the arrangement had a significant effect on the number of fixations, where top-to-bottom arrangement brought a higher number of fixations. The background had a significant effect on accuracy and subjective visual fatigue, where the accuracy and subjective visual fatigue without a background was better than with a background. For the analysis of the scan path, a denser scan path was found in text than in animation, in top-to-bottom arrangement than in left-to-right arrangement, and without a background than with a background. In the future, game manufacturers should use the results of this research to design different “Spot the Difference” videos. When designing a simple game, an animation without a background and involving a left-to-right arrangement was recommended. When designing a difficult game, the opposite settings should be used.

Scalable Thermal Simulation of Powder Bed Fusion

Powder bed fusion (PBF) has become a widely used additive manufacturing technology to produce metallic parts. In PBF, thermal field evolution during the manufacturing process plays an important role in determining both geometric and mechanical properties of the fabricated parts. Thermal simulation of the PBF process is computationally challenging due to the geometric complexity of the manufacturing process and the inherent computational complexity that requires a numerical solution at every time increment of the process. We propose a new thermal simulation of the PBF process based on the laser scan path. Our approach is unique in that it simulates the thermal history of the process on the discretization of the geometry implied by the process plan, as opposed to voxelization or meshing of the design geometry. The discretization is based on the laser scan path, and the thermal model is formulated directly in terms of the manufacturing primitives. An element growth mechanism is introduced to simulate the evolution of the melt pool during the manufacturing process. A spatial data structure, called contact graph, is used to represent the discretized domain and capture all thermal interactions. The simulation is localized through exploiting spatial and temporal locality. This limits the need to update to at most a constant number of elements at each time step, which implies that the proposed simulation not only scales to handle 3D components of arbitrary complexity but also can achieve real-time performance. The simulation is fully implemented and validated against experimental data and other simulation results.