A k-means implementation that clusters RGB values of a given image and then writes the new values to an image file.
At first, all the RGB values are sent to Processor actor to start clustering. Processor selects k centroids randomly and spawns a ReducerActor for each. Processor sends RGBs to mapperRouter which is a round-robin router of MapperActors. Each MapperActor finds the closest centroid and send RGB data to associated ReducerActor. Once all the data are processed, the first iteration finishes. This process continues until the totalIteration is reached. At last, a new image is created according to final centroid values.
| Original Image | Clustered Image |
|---|---|
 |
 |
There is an example image in image directory. It is taken from the movie WALL-E. By default, image is the input/output directory of the program. You can change it by specifying a different imagePath in Application object. To run the program type: sbt run