A Delaunay triangulation based approach for cleaning rough sketches

Given a set of rough strokes drawn by an artist (either in pen-paper medium or in digital medium) in raster format, the objective is to group them meaningfully and represent the group with simple most appropriate curves. In this paper, a Delaunay triangulation based algorithm is proposed for grouping strokes. The grouping procedure is capable of identifying open curves and reconstructing broken strokes. The proposed algorithm is capable of helping the user in masking misinterpreted regions. We also introduce a shape aware skeleton smoothing procedure which best approximates the shape by taking input raster sketch as a reference to create final vector output. The user can also control the final output. The proposed algorithm combines the techniques in computational geometry as well as in image processing to utilize the power of both.