Add BLOG.md

Author: PascalDierich

.gitignore

@@ -1,3 +1,7 @@
 .idea
 vendor/
-image-clustering
+
+# binaries
+image-clustering
+bin/
+current

BLOG.md

@@ -0,0 +1,26 @@
+# Image clustering in Go
+
+The _image-clustering_ program partitions the pixel in an image
+based on their color to a pre-defined number of clusters.
+Each cluster is a random color assigned. The resulting image 
+will represent each pixel in its corresponding cluster-color.
+
+
+#### Example
+
+original         |  result with 10 clusters
+:-------------------------:|:-------------------------:
+![](data/img3.png)  	|  ![](data/img3_c10.png)
+
+
+### K-Means Clustering
+
+To achieve this we use __K-Means Clustering__, an iterative algorithm to partition a set
+of data points (here colors) into a pre-defined number (k) of clusters.
+
+The algorithm starts by assigning each cluster a randomly created centroid.
+The centroid is the _"average"_ of the points in the cluster.<br>
+After we assigned each pixel based on their color
+to the cluster with nearest centroid, we recompute the centroids 
+by assigning them to the average of their members.
+This two steps are repeated until the centroids converged.