Line Labeling Using Markov Random Fields

The task of obtaining a line labeling from a greyscale image of trihedral objects presents diiculties not found in the classical line labeling problem. As originally formulated, the line labeling problem assumed that each junction was correctly pre-classiied as being of a particular junction type (e.g. T, Y, arrow); the success of the algorithms proposed have depended critically upon getting this initial junction clas-siication correct. In real images, however, junctions of diierent types may actually look quite similar, and this pre-classiication is often diicult to achieve. This issue is addressed by recasting the line labeling problem in terms of a coupled probabilis-tic system which labels both lines and junctions. This results in a robust system, in which prior knowledge of acceptable conngurations can serve to overcome the problem of misleading or ambiguous evidence. suggested the application of MRFs to the line labeling problem.