| Summary: | 碩士 === 國立暨南國際大學 === 資訊工程學系 === 103 === With the advance of science and technology, mobile devices become popular products. In this thesis, we propose to apply shape recognition technology to realize an interactive sketch learning system. In this system, it offers the learner different sketch samples that the learner can look at the sample and imitate sketch drawing. Once the learner finishes, the system will feedback a score that is computed from the shape similarity. Users can earn self-affirmation and feel fulfilled in the system thus help users to continue learning sketch.
The following shape features are extracted from the sketches: 2-D Fourier descriptor, shape context and TPS-RPM。The similarity between the sample and learner’s sketch is computed to give the user a grading score. To realize on mobile platforms, we conduct experiments to find the most efficient and accurate features to build an accurate and fast sketch scoring system. In order to evaluate our scoring mechanism, subjective scores of users’ drawings are evaluated to analyze whether the system score is close to people's subjective feelings. Experiments include 6 courses, where there are 8 users’ drawings in 5 courses and 6 users’ drawings in the reverse Picasso course. Analysis of these courses using linear regression, the highest R square is 0.9673.
|
|---|
