WIP: Demo on ImageSegmentation.jl Utilities tools

docs/examples/image_segmentation/rag.jl

@@ -0,0 +1,73 @@
+# ---
+# cover: assets/rag.gif
+# title: Region Adjacency Graph
+# description: This demo shows how to use the region_adjacency_graph
+# author: Ashwani Rathee
+# date: 2021-09-5
+# ---
+
+# In this demonstraton, we will study about region adjacency graph and region tree.
+
+# Region Adjacency Graph (RAG), $G=V,E$, is composed of nodes/vertices representing the regions and edges representing the adjacency.
+# Region Adjacency Graphs, as the name suggests represent adjacency of regions
+# with a graph. Each region in the image is a node in a graph.
+
+# There is an edge between every pair of adjacent regions (regions whose pixels are 
+# adjacent). The weight of between every two nodes can be defined in a variety of ways.
+# For this example, we will use the difference of average color between two regions as
+# their edge weight. The more similar the regions, the lesser the weight between
+# them. Because we are using difference in mean color to compute the edge weight,
+# the method has been named rag_mean_color.
+
+# # Creating a Region Adjacency Graph (RAG)
+
+# A region adjacency graph can directly be constructed from a SegmentedImage
+# using the `region_adjacency_graph` function. Each segment is denoted by a 
+# vertex and edges are constructed between adjacent segments. The output 
+# is a tuple of SimpleWeightedGraph and a Dict(label=>vertex) with weights assigned according to weight_fn.
+
+using Images
+using ImageSegmentation, Distances, TestImages
+
+# We will use coffee image for this demonstration.
+
+img = testimage("coffee")
+seg = felzenszwalb(img, 10, 100)
+
+# Here we calculate the weight of the graph edges using euclidean distance between 
+# 2 segments.
+
+weight_fn(i,j) = euclidean(segment_pixel_count(seg, i), segment_pixel_count(seg, j))
+
+# `region_adjacency_graph` requires the seg which has provides label map and weight to
+# create the graph.
+
+G, vert_map = region_adjacency_graph(seg, weight_fn)
+
+# G is an undirected simple Int64 graph with Float64 weights. `region_adjacency_graph`
+# returns the RAG along with a Dict(label=>vertex) map.
+
+# Here, the difference in pixel count has been used as the weight 
+# of the connecting edges. This difference measure can be useful 
+# if one wants to use this region adjacency graph to remove smaller
+# segments by merging them with their neighbouring largest segment.
+# Another useful difference measure is the euclidean distance 
+# between the mean intensities of the two segments.
+
+
+# ## Creating a Region Tree
+
+# A region tree can be constructed from an image using `region_tree` function. 
+# If the image is not homogeneous, then it is split into half along each 
+# dimension and the function is called recursively for each portion of the 
+# image. The output is a RegionTree.
+
+function homogeneous(img)
+    min, max = extrema(img)
+    max - min < 0.2
+end
+
+t = region_tree( Gray.(img), homogeneous)        # `img` is an image
+
+
+save("assets/rag.gif", cat(img, Gray.(img); dims=3); fps=1) #src