Details

Type

  • Bachelor Thesis
  • Student Project
  • Master Thesis

Persons

1-2
Workgroup
Rendering and Modeling

Motivation

Several Apps exist for scanning an object or an interior room with your mobile to produce a 3D model, e.g., RealityScan, but there usually is no feedback on the completeness or quality of the scan. Augmented reality guides such as arrows pointing to the Next Best View (NBV) estimated from the continuously updating input scan, text labels with information, and a point cloud overlay with a quality indicator can guide the user to efficiently acquire a complete scan with the desired quality of details.

Description

Extend the existing SLAM framework RTABMap to display suitable Augmented Reality graphic guides from analysis of the input scans (missing data, scan quality) and validate its effectiveness with user studies, on a mobile device by communicating with a server or directly on the mobile device.

Tasks (depending on PR/BA/DA and number of students)

  • Analyze the Next Best View (NBV) based on the currently scanned point cloud using a network
  • Segment the object from the background so that NBV yields proper results
  • Adapt the NBV network to work in an environment scene instead of just circling the sensor around an object (inverse and more complex perspective)
  • Display graphical elements in Augmented Reality such as arrows, text labels, and a point cloud overlay indicating its quality
  • Analyze scan quality, e.g., based on sampling density
  • Estimate scan completeness
  • Output a robust and good quality 3D mesh from the scanned point cloud
  • Evaluation of the effectiveness of the visual guides with a user study (scan time, scan quality)
  • Port algorithms from Kinect Azure sensor/PC onto a smartphone (either all natively, or only scan input+AR display, connecting to a server for the rest)

Requirements

  • Knowledge of English (source code comments and final report have to be in English)
  • Knowledge of basic geometry processing and C++ is required, experience in mobile coding (Fluttr/Kotlin) is a plus since all algorithms have to run on a smartphone

Environment

The project should be implemented platform independently (Android, iOS).

A bonus of €500/€1000 if completed to satisfaction within an agreed time-frame of 6/12 months (PR/BA or DA)

Responsible

For more information please contact Stefan Ohrhallinger.