image-segmentation

Try it online

Image segmentation using Minimum Spanning Forests.

Implements:

• Efficient Graph-Based Image Segmentation (GBS)
• An Efficient Parallel Algorithm for Graph-Based Image Segmentation (PHMSF)

Also, as a helper for the second segmentation algorithm:

• Block-based noise estimation using adaptive Gaussian filtering

Developed as a project for the course "Graph Theory, Networks and Applications" at Mälardalen University. The course's project report can be found here.

Setup

Prerequisites:

• Go
• Node
• Bower

$ go get github.com/miguelfrde/image-segmentation
$ cd $GODIR/src/github.com/miguelfrde/image-segmentation
$ bower install
$ mkdir tmp

The tmp directory needs to be created since result images will be served from it in Heroku.

Run

$ PORT=8080 go run main.go

Go to localhost:8080 and try some images!

Test

$ go test ./...

TODO

• Implement the parallel version of the PHMSF algorithm
• Implement more MST or non-graph based segmentation algorithms