In this section, we discuss how the Region Adjacency Graph (RAG) can be used to combine the over-segmented regions of an image, to obtain a better segmentation. It uses the SLIC algorithm to first segment the input image and obtain the region labels. Then, it constructs a RAG and progressively merges the over-segmented regions that are similar in color. Merging two adjacent regions produces a new region with all the pixels from the merged regions. Regions are merged until no highly similar region pairs remain:

from skimage import segmentation
from skimage.future import graph

def _weight_mean_color(graph, src, dst, n):
 diff = graph.node[dst]['mean color'] - graph.node[n]['mean color']
 diff = np.linalg.norm(diff)
 return {'weight': diff}

def merge_mean_color(graph, src, dst):
 graph.node[dst]['total color'] += graph.node[src]['total color']
 graph.node[dst]['pixel count'] += graph.node[src]['pixel count']
 graph.node[dst]['mean color'] = (graph.node[dst]['total color'] /  graph.node[dst]['pixel count'])

img = imread('../images/me12.jpg') 
labels = segmentation.slic(img, compactness=30, n_segments=400)
g = graph.rag_mean_color(img, labels)
labels2 = graph.merge_hierarchical(labels, g, thresh=35, rag_copy=False,
 in_place_merge=True,
 merge_func=merge_mean_color,
 weight_func=_weight_mean_color)
out = label2rgb(labels2, img, kind='avg')
out = segmentation.mark_boundaries(out, labels2, (0, 0, 0))
pylab.figure(figsize=(20,10))
pylab.subplot(121), pylab.imshow(img), pylab.axis('off')
pylab.subplot(122), pylab.imshow(out), pylab.axis('off')
pylab.tight_layout(), pylab.show()

The next screenshot shows the output of the code, the input image, and the output image obtained using RAG, merging the SLIC segments: