Recovery of drawing order from multi-stroke English handwritten images based on graph models and ambiguous zone analysis

Recovery of the drawing order of strokes in a handwritten image can be seen as searching for the smoothest path for each stroke on an undirected graph that is constructed from the skeleton of the handwritten image. However, this requires correcting for separating strokes, and detecting starting points. Moreover, ambiguousness at junction points increases the complexity of finding the smoothest paths. In order to resolve these issues, an effective approach that can simultaneously detect the points to separate strokes and find the optimal path for each stroke is proposed. To reduce the complexity of the problem, the skeleton graph of the handwritten image is used, and touching characters or crossing strokes are separated. Touches or crossings of stroke parts at ambiguous zones are detected and the smoothness values are adjusted to improve the accuracy. The greedy algorithm and Dijkstra'salgorithm with a well-defined function of smoothness are applied in searching the optimal path. The nature of the recovery is increased when the optimal path is split into many strokes by using the curvatures of the edges, the un-smoothness between edges and the appearance of double-traced edges. Finally, pixel sequences of strokes are extracted and ordered by using rules of handwriting. The effectiveness of the proposed method is demonstrated through low error rates of pixel sequence comparison and high accuracy of online recognition.