Citation:
Hammond, Tracy, and Brandon Paulson. "Recognizing sketched multistroke primitives." ACM Transactions on Interactive Intelligent Systems (TiiS) 1.1 (2011): 4
Summary:
Most research in primitive recognition often assumes that a basic shape such as lines and polygons are are drawn with a single stroke. If multi strokes are allowed, the user is forced to draw under certain set of limitations. This paper seeks to eliminate, by giving the user full independence to complete one shape in multiple strokes, or multiple shapes in a single stroke. The main stages in this multi stroke recognition system are, graph building, identification of strongly connected components, merging components and removing false positives.
Discussion:
By analyzing the sketches of a set of users over a range of shapes, it was found that there were a lot differences in the styles of drawing, especially as far as number of strokes in a shape are concerned. Many people tend to make use of touch up strokes and continuity strokes, even while drawing a simple figure.
1. Use the stroke information to form a graph. The endpoints of strokes form th nodes, while an edge between two nodes occurs if either of the three conditions are satisifed:
-There is a stroke between the two points
- The gap between the two points is within a threshold of the average stroke length of the strokes passing through the end point
- The gap is smaller than the average width of the bounding box of the two strokes.
2. Tarjan's algorithm is used to extract the strongly connected components in this graph.
3. The merging of nodes identified done in the same component is done by removing peaks at the end points, and finding the point in the two strokes closest to each other, and creating a touch up stroke between them.
4. The final step is remove false positives.
Discussion:
By analyzing the sketches of a set of users over a range of shapes, it was found that there were a lot differences in the styles of drawing, especially as far as number of strokes in a shape are concerned. Many people tend to make use of touch up strokes and continuity strokes, even while drawing a simple figure.
1. Use the stroke information to form a graph. The endpoints of strokes form th nodes, while an edge between two nodes occurs if either of the three conditions are satisifed:
-There is a stroke between the two points
- The gap between the two points is within a threshold of the average stroke length of the strokes passing through the end point
- The gap is smaller than the average width of the bounding box of the two strokes.
2. Tarjan's algorithm is used to extract the strongly connected components in this graph.
3. The merging of nodes identified done in the same component is done by removing peaks at the end points, and finding the point in the two strokes closest to each other, and creating a touch up stroke between them.
4. The final step is remove false positives.
No comments:
Post a Comment