Image Segmentation Demo

Implementation of four classical image segmentation methods in C compiled for the web.

Usage

Go to https://inventshah.github.io/image-segmentation/ in your browser.

Or run locally:

1. git clone https://github.com/inventshah/image-segmentation.git
2. Start a static server python -m http.server 8080
3. Go to localhost:8080 in your browser.

Change the parameters for each algorithm, toggle gamma correction, or click on an image to rerun the algorithm.

Web Build

The clang compiler must be used.

Run make wasm

Non-make users: clang -Os -flto -msimd128 -mbulk-memory --target=wasm32 -nostdlib -Wl,--no-entry -Wl,--allow-undefined -o main.wasm src/meanshift.c src/kmeans.c src/histogram.c src/splitmerge.c src/utils.c src/utils_wasm.c

Native Build

Download the libpng dependency.

Run make native

Non-make users: cc -O3 -flto -o segment -lpng src/native.c src/kmeans.c src/meanshift.c src/histogram.c src/splitmerge.c src/utils.c

Run any of the following to produce an out.png file:

./segment kmeans <source png> <k value>
./segment meanshift <source png> <bandwidth value>
./segment splitmerge <source png> <tolerance>
./segment histogram <source png>

Individual repositories

The implementations of the algorithms in this repository are based on the following:

• https://github.com/inventshah/Kmeans-C
• https://github.com/inventshah/Mean-Shift-C
• https://github.com/inventshah/Balanced-Histogram-Thresholding-C
• https://github.com/inventshah/Split-Merge-C

This repository updates those implementations for better WASM support.