الفهرس Only 14 pages are availabe for public view
 |  | from | 93 |  |  |
| | | from | 93 | | |
Abstract
With the recent advances in tablet devices industry, sketch recognition has become a potential replacement for existing systems’ traditional user interfaces. Structured dia- grams (flow charts, markov chains, module dependency diagrams, state diagrams, block diagrams, UML, graphs, etc.) are very common in many science fields. Usually, such diagrams are created using structured graphics editors like microsoft visio. Structured graphics editors are extremely powerful and expressive, but they can be cumbersome to use. This thesis presents a domain-independent sketch recognition system that con- verts user’s sketch into structured shapes in usable electronic format with minimal ef- fort. We have presented a robust grouping technique which is based on spatial distance between strokes. We have also introduced the concept of classification preprocessing which is proved to minimize classifiers confusion by doing a precategorization to the shapes. We have compared the accuracies of three classifiers; support vector machines (SVMs), random forest (RF), and K- Nearest neighbor K - NN