https://github.com/rylanschaeffer/droplet-analysis

2015 Senior Design Project
https://github.com/rylanschaeffer/droplet-analysis

Last synced: 3 days ago
JSON representation

2015 Senior Design Project

• Host: GitHub
• URL: https://github.com/rylanschaeffer/droplet-analysis
• Owner: RylanSchaeffer
• Created: 2015-07-17T01:17:59.000Z (over 9 years ago)
• Default Branch: master
• Last Pushed: 2015-07-21T05:51:30.000Z (over 9 years ago)
• Last Synced: 2024-10-12T21:13:08.403Z (about 1 month ago)
• Language: Matlab
• Size: 1.87 MB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Droplet-Analysis

For ECS193 (Senior Design), Ramya Bhaskar, Amanda Ho, Willie Huey and I completed a project for Professor Benjamin Shaw (http://faculty.engineering.ucdavis.edu/shaw/). The project was to design and implement an algorithm that could (with relative accuracy) predict the diameter of a fuel droplet combusting in a zero-gravity environment. Although the droplets are roughly circular, the problem is compounded by the presence of "soot," which adds noise to the image. Five simulated image sequences are publicly available at https://drive.google.com/folderview?id=0BxVyAsYY7QlqfjhJaWcwMVRaUHhvOWNvX3pfa0Q0VTgwRmg2MWQ2M3l0UUw1cU0xdGFiMms&usp=sharing. The file "droplet size vs frame number.csv" (located in the same folder as this README) contains the true diameter means used to generate the images.

Amanda, Willie and I worked on a MATLAB implementation, while Ramya explored OpenCV. Our first approach was to use built in MATLAB functions and basic statistical techniques to predict "circles of best fit" for each image. These scripts are in the folder titled "IMFindCircles," based on the main function used.

Hopefully, I'll have time to explore some machine learning algorithms (likely a convolution neural net) to see what improvements they might offer over the first approach.