Presentation Title

Digital 3D Image Reconstruction via Sonar Rangefinding

Faculty Mentor

Craig Reinhart Ph.D.

Start Date

17-11-2018 8:30 AM

End Date

17-11-2018 10:30 AM

Location

HARBESON 65

Session

POSTER 1

Type of Presentation

Poster

Subject Area

engineering_computer_science

Abstract

3D image reconstructionis the process by which a 3D model is generated using a set of 2D images. One of the main challenges of image reconstruction is depth determination; projecting a 3D scene onto a 2D image results in loss of depth, an essential visual component of 3D objects. Much of work in image reconstruction focuses on the techniques and algorithms that infer depth from images. Specifically, many methods attempt to model the human visual system and utilize visual cues, such as stereo parallax, occlusion, and focus. However, a different approach for depth determination is rangefinding, which utilizes time-of-flight sensors for distance measurements. Rather than inferring depth information from images, depth data can be directly measured using time-of-flight sensors. The focus of this project was to create a 3D image reconstruction system that measures depth in real-time using a sonar rangefinder.

Summary of research results to be presented

The reconstruction system used the following tools: a robotic arm, a digital camera, a sonar sensor. The camera and sonar sensor were mounted to the robotic arm and calibrated to enable accurate measurement of scene data. Thereafter, the camera and sonar sensor were used to collect image and depth data from the scene. The robotic arm was used to allow both sensors to accurately move both sensors through space. After data collection, image and depth data were registered in a single coordinate system and visually rendered into a 3D representation of the scene using the image processing software “ImageJ”. Multiple experiments were conducted to find the optimal parameters for image and depth data collection. The implementation of the image reconstruction system was successful. Results show that the system can create 3D models of primitive shapes, such as rectangular prisms. Future work will be done to measure its performance on more complex shapes. The sonar sensor is limited due to surface reflectance variability and noise and, thus, exploring other sensors, such as LiDAR, may be worthwhile.

This document is currently not available here.

Share

COinS
 
Nov 17th, 8:30 AM Nov 17th, 10:30 AM

Digital 3D Image Reconstruction via Sonar Rangefinding

HARBESON 65

3D image reconstructionis the process by which a 3D model is generated using a set of 2D images. One of the main challenges of image reconstruction is depth determination; projecting a 3D scene onto a 2D image results in loss of depth, an essential visual component of 3D objects. Much of work in image reconstruction focuses on the techniques and algorithms that infer depth from images. Specifically, many methods attempt to model the human visual system and utilize visual cues, such as stereo parallax, occlusion, and focus. However, a different approach for depth determination is rangefinding, which utilizes time-of-flight sensors for distance measurements. Rather than inferring depth information from images, depth data can be directly measured using time-of-flight sensors. The focus of this project was to create a 3D image reconstruction system that measures depth in real-time using a sonar rangefinder.