MaRoboticCTTargetSegmentation

Chair for Computer Aided Medical Procedures & Augmented Reality
Lehrstuhl für Informatikanwendungen in der Medizin & Augmented Reality

Thesis: Robotic Computed Tomography: Image segmentation for target tracking

Thesis by:
Advisor: Tobias Lasser
Supervision by: Erdal Pekel

Abstract

In Robotic Computed Tomography a robot places a target object in different poses between source and detector for improved reconstruction results. Precise positioning of the robot's end-effector is critical for the reconstruction of the target object. In this thesis a real-time, GPU-based (RGB) image segmentation algorithm will be implemented in OpenCL. It will enable the analysis of the target object's positioning error relative to the detector's center.

Tasks

The aim of this thesis is the implementation of a real-time, (GPU-based) RGB image segmentation algorithm that analyses the target object's positioning error relative to the detector's center.

The following tasks need to be carried out throughout the duration of the project:

  • Retrieval and decoding of the camera image via network interface
  • Image preprocessing
  • Consideration of system state for segmentation task (robot state, target type and dimensions, etc.)
  • Custom GPU-based segmentation algorithm implementation in OpenCL
  • Integration of the algorithm within existing ROS package after successful proof of concept

Requirements

Computer Vision Algorithms, C++, GPU Programming, Interest in Robotics

Contact


Students.ProjectForm
Title: Robotic Computed Tomography: Image segmentation for target tracking
Abstract: In Robotic Computed Tomography a robot places a target object in different poses between source and detector for improved reconstruction results. Precise positioning of the robot's end-effector is critical for the reconstruction of the target object. In this thesis a real-time, GPU-based (RGB) image segmentation algorithm will be implemented in OpenCL. It will enable the analysis of the target object's positioning error relative to the detector's center.
Student:  
Director: Tobias Lasser
Supervisor: Erdal Pekel
Type: Master Thesis
Area:  
Status: open
Start: 01.10.2019
Finish:  
Thesis (optional):  
Picture:  


Edit | Attach | Refresh | Diffs | More | Revision r1.5 - 19 Aug 2019 - 15:53 - TobiasLasser