ImageCompressor

A nice application to parallel k-means

Prerequisites

• Haskell
• Stack
• Ghcup
• Make

Description

Second year Epitech project made in Haskell. The goal is to reduce the size of an image by implementing the K-means algorithm. Three steps are needed for the implementation:

• read the image and extract the colors of each pixel,
• cluster these colors, and replace each color of a given cluster by the mean color of this cluster,
• index the means of the cluster, and create the compressed image. The project focuses on the second part, that is clustering.

Clustering

Arguments

The following arguments are required for the program:

Option	Description
-n	number of colors in the final image
-l	convergence limit
-f	path to the file containing the colors of the pixels

Grammar

• The list of pixel given in the file passed as argument follow the following grammar:

IN      ::= POINT ' ' COLOR ( '\n' POINT ' ' COLOR )*
POINT   ::= '(' int ',' int ')'
COLOR   ::= '(' SHORT ',' SHORT ',' SHORT ')'
SHORT   ::= '0'.. '255'

• The output, which is the clustered color, follow this grammar:

OUT     ::= CLUSTER*
CLUSTER ::= '- -\n' COLOR '\n -\ n' ( POINT ' ' COLOR '\n')*
POINT   ::= '(' int ',' int ')'
COLOR   ::= '(' SHORT ',' SHORT ',' SHORT ')'
SHORT   ::= '0 '.. '255'

Build program

To compile the imageCompressor, stack is used wrapped in a Makefile, which implements the following rules:

Command	Result
make	create a imageCompressor executable with stack build. Then copy the produced executable at the root of the project.
make clean	delete the files produced by the compilation (in that case execute stack clean).
make fclean	execute a stack purge, and then delete the imageCompressor executable at the root of project.
make re	do the fclean and all rules (in that order)
make tests_run	execute unit_tests with stack test, and copy the coverage in test/coverage/ folder.

Authors

• Luka Camus
• Louis Bassagal