Create a region adjacency graph (rag) of a vertex labelled graph. Create a region adjacency graph (rag) from a graph cut. Projects rag vertex weights onto original graph vertices. rag_back_project_edge_weights(graph, …) Projects rag edge weights onto original graph edges. rag_accumulate_on_vertices(rag, accumulator, …) Weights rag vertices by accumulating values from the vertex weights of the original graph. rag_accumulate_on_edges(rag, accumulator, …) Weights rag edges by accumulating values from the edge weights of the original graph.
make_region_adjacency_graph_from_labelisation(graph, vertex_labels)[source]

Create a region adjacency graph (rag) of a vertex labelled graph. Each maximal connected set of vertices having the same label is a region. Each region is represented by a vertex in the rag. There is an edge between two regions of labels $$l_1$$ and $$l_2$$ in the rag iff there exists an edge linking two vertices of labels $$l_1$$ and $$l_2$$ int he original graph.

Parameters:
• graph – input graph

• vertex_labels – vertex labels on the input graph

Returns:

a region adjacency graph (Concept CptRegionAdjacencyGraph)

make_region_adjacency_graph_from_graph_cut(graph, edge_weights)[source]

Create a region adjacency graph (rag) from a graph cut. Two vertices $$v_1$$, $$v_2$$ are in the same region if there exists a $$v_1v_2$$-path composed of edges weighted 0. Each region is represented by a vertex in the rag. There is an edge between two regions of labels $$l_1$$ and $$l_2$$ in the rag iff there exists an edge linking two vertices of labels $$l_1$$ and $$l_2$$ int he original graph.

Parameters:
• graph – input graph

• edge_weights – edge weights on the input graph, non zero weights are part of the cut

Returns:

a region adjacency graph (Concept CptRegionAdjacencyGraph)

rag_back_project_vertex_weights(graph, vertex_weights)[source]

Projects rag vertex weights onto original graph vertices. The result is an array weighting the vertices of the original graph of the rag such that: for any vertex $$v$$ of the original graph, its weight is equal to the weight of the vertex of the rag that represents the region that contains $$v$$.

For any vertex index $$i$$, $$result[i] = rag\_vertex\_weight[rag\_vertex_map[i]]$$

Parameters:
• graph – input region adjacency graph

• vertex_weights – vertex weights on the input region adjacency graph

Returns:

vertex weights on the original graph

rag_back_project_edge_weights(graph, edge_weights)[source]

Projects rag edge weights onto original graph edges. The result is an array weighting the edges of the original graph of the rag such that: for any edge $$e$$ of the original graph, its weight is equal to the weight of the edge of the rag that represents that links the two regions containing the extremities of $$e$$. If no such edge exists (if the extremities of $$e$$ are in the same region), its value is 0.

For any edge index $$ei$$, $$result[ei] = rag\_edge\_weight[rag\_edge\_map[ei]]$$ if $$rag\_edge\_map[ei] != -1$$ and 0 otherwise.

Parameters:
• graph – input region adjacency graph

• edge_weights – edge weights on the input region adjacency graph

Returns:

edge weights on the original graph

rag_accumulate_on_vertices(rag, accumulator, vertex_weights)[source]

Weights rag vertices by accumulating values from the vertex weights of the original graph.

For any vertex index $$i$$ of the rag, $$result[i] = accumulator(\{vertex\_weights[j] | rag\_vertex\_map[j] == i\})$$

Parameters:
• rag – input region adjacency graph (Concept RegionAdjacencyGraph)

• vertex_weights – vertex weights on the original graph

• accumulator – see Accumulators

Returns:

vertex weights on the region adjacency graph

rag_accumulate_on_edges(rag, accumulator, edge_weights)[source]

Weights rag edges by accumulating values from the edge weights of the original graph.

For any edge index $$ei$$ of the rag, $$result[ei] = accumulate(\{edge\_weights[j] | rag\_edge\_map[j] == ei\})$$

Parameters:
• rag – input region adjacency graph (Concept RegionAdjacencyGraph)

• edge_weights – edge weights on the original graph

• accumulator – see Accumulators

Returns:

edge weights on the region adjacency graph