This page provides an overview of the problem that motivates this project and the strategy to solve it. It also provides a timeline for selecting a stage and seeing more details of that stage.

Title: Identification techniques of operational vibration modes with MotionScope assisted by stereoscopy

Título: Técnicas de identificação de formas operacionais de vibração com motionscope assistido por estereoscopia

Problem description

The techniques currently available to identify operational vibration modes using video cameras rely only on one viewpoint. The problem is that the results could be badly interpreted with only one point of view since certain vibration modes look similar to others when projected to the camera's plane, causing “optical illusions.” The following GIF is a section of a motionscope output video.

Source:https://www.motion-scope.com/motion-magnification?pgid=l5c7rsyb-2aa8b6a7-1a9d-4429-9601-24bb2aca6d4a

In this image, it is possible to see that, if we didn't know what to expect, it would be tough to guess if the drone's arm is moving right and left or back and forth relative to the image plane.

Constraints

• To avoid possible inconveniences and increase flexibility of use, avoid placing the camera in predefined locations or measuring its positions.
• The zoom and focus should be adjustable from position to position.
• The lighting conditions are constant.

Objectives

The main goal of this project is to develop a system that can be parallel or integrated into MotionScope and allows the combination of two or more viewpoints to obtain the 3D displacement of each point so that it is possible to determine the vibration mode with certainty. To do this, it is necessary to accomplish the following objectives:

• Implementing or developing an algorithm to find corresponding points between images and track them
• Finding the intrinsic and extrinsic parameters of the camera(s) using the information about the corresponding points OR an alternative method that allows to determine the relative position of each camera
• Merging the equivalent points to get the 3D position and displacement
• Developing a tool to watch the results (maybe an animated 3D plot)
• (Optional) Wrap the points around a 3D model for a better interpretation of the results

Milestones