Skip to content

Visualization of powerful boundary detection tools

nasyring edited this page Mar 24, 2016 · 20 revisions

Background

Detecting boundary of an image based on noisy observations is a fundamental problem of image processing and image segmentation. It has wide applications such as tumor detection in medical fields, development assessment in economics, climate changes in environment studies, etc. While there are many matured statistical methods in this field, the visualization of them is far from satisfactory thus prevents a larger scientific community to actually use them. This project will improve the visualization of the state-of-the-art statistical boundary detection methods, considering subject matter fields for the implementation.

Related work

There are not many R packages on boundary detection. As an exception, the R package `BayesBD` is a package to implement nonparametric Bayesian boundary detection, however, there is a lot of room to improve the speed of the package and the visualization is lacking.

Details of your coding project

The main reference is the paper in the Reference. The potential students are expected to write efficient code in C/C++ to improve the associated R package `BayesBD`. There are two main functions: `BayesBD` to turn noisy observations to detected boundary, and `visualBD` to visualize the estimation procedure. Testing and documentation are of course required to help popularize this approach. In addition, students can find other competing approaches to make this package more comprehensive

Expected impact

Visualization of an advanced statistical method is critical to make the “outsiders” to actually use it, which also popularizes the usage of R. By using the existing package as a benchmark, this will be an encouraging story of how a professional improvement can bring to the R community.

Mentors

Please contact Meng Li <[email protected]> and Vijay Barve http://vijaybarve.net/.

Tests

  • Medium: write a C/C++ function for a loop and call this function in R.

Solutions of tests

Students, please post a link to your test results here.

Abhinav Agarwal repo

Rohan test_BayesBD

Mainak Mandal test_code_BayesBS

Nicholas Syring BayesBD Test

Reference

Li, M. and Ghosal, S.(2015). Bayesian Detection of Image Boundaries.

Li, M. (2015), BayesBD, R package for Bayesian boundary detection in images using Gaussian process priors.

Clone this wiki locally