K-means and DBSCAN

K-means and DBSCAN are clustering algorithms, which we apply for color segmentation in images.

K-means tries to find a color representatives for a number of classes given, i.e., most average color for each class, which is most similar to the colors within the class but as different as possible from colors in other classes.

DBSCAN is so called density-based clustering algorithm, which tries to group similar colors into different classes based on how densely they are positioned together.

Files

0_read.py - reads image and outputs its pixel values as vector array

0_kmeans.py - reads vector array, performs k-means and outputs vector array

0_dbscan.py - reads vector array, performs dbscan and outputs vector array

0_save.py - reads vector array and generates an image

log.txt - contains clustering information for pictures

Dependencies

All packages/libraries should be included in Python 3.6.8

Limitations

Image generation with 0_save.py works only for 3 or 4 channel models i.e. RGB (.jpg) or RGBA (.png)

Format

k-means and dbscan use following format string (string of 3 dimensional array of size [x,y,n]):

[[[n0,n1,n2,...],[n0,n1,n2,...]],[[n0,n1,n2,...],[n0,n1,n2,...]],...]

Usage

k-means

python3 0_read.py <inputImageName> | python3 0_kmeans.py <k-value> | python3 0_save.py <outputImageName>

dbscan

Parameter "distanceFunction" uses only "e" for euclidean distance or "m" for maximum distance

python3 0_read.py <inputImageName> | python3 0_dbscan.py <eps> <minPts> <distanceFunction> | python3 0_save.py <outputImageName>

Examples

k-means

python3 0_read.py 1_butterfly.jpg | python3 0_kmeans.py 2 | python3 0_save.py 2_butterfly2means.jpg
python3 0_read.py 1_butterfly.jpg | python3 0_kmeans.py 5 | python3 0_save.py 2_butterfly5means.jpg

python3 0_read.py 1_pug.jpg | python3 0_kmeans.py 2 | python3 0_save.py 2_pug2means.jpg
python3 0_read.py 1_pug.jpg | python3 0_kmeans.py 5 | python3 0_save.py 2_pug5means.jpg

python3 0_read.py 1_cat.jpg | python3 0_kmeans.py 2 | python3 0_save.py 2_cat2means.jpg
python3 0_read.py 1_cat.jpg | python3 0_kmeans.py 5 | python3 0_save.py 2_cat5means.jpg

dbscan

python3 0_read.py 1_butterflySmaller.jpg | python3 0_dbscan.py 100 100 e | python3 0_save.py 2_butterfly_dbscan_100_100_e.jpg
python3 0_read.py 1_butterflySmaller.jpg | python3 0_dbscan.py 100 100 m | python3 0_save.py 2_butterfly_dbscan_100_100_m.jpg

python3 0_read.py 1_pugSmaller.jpg | python3 0_dbscan.py 100 100 e | python3 0_save.py 2_pugSmaller_dbscan_100_100_e.jpg
python3 0_read.py 1_pugSmaller.jpg | python3 0_dbscan.py 100 100 m | python3 0_save.py 2_pugSmaller_dbscan_100_100_m.jpg

python3 0_read.py 1_catSmaller.jpg | python3 0_dbscan.py 100 100 e | python3 0_save.py 2_catSmaller_dbscan_100_100_e.jpg
python3 0_read.py 1_catSmaller.jpg | python3 0_dbscan.py 100 100 m | python3 0_save.py 2_catSmaller_dbscan_100_100_m.jpg

Copyright

All the images were taken from internet and therefore belong to their respective owners.

Algorithms were not invented by me, I merely just implemented them.