Study Of The Handwritten Sketch Recognition Algorithm

With the popularization of touch technology and related equipment, the handwritten input has become one of the most natural human-computer interaction form, especially in the field of preschool education, it has been widely used. Sketch is an important form of drawing, the features of which are simple abstract, vivid, strong generalization and arbitrary. Handwritten sketch can inspire and develop children’s creativity and imagination, it can also help improve their expression ability. So handwritten sketch intelligent recognition is of great importance to the promotion and application of sketch, In addition, intelligent recognition of handwritten graphics object is also the important development direction of the digital ink technology.Based on the analysis of the sketch composition of graphic element, the paper generalize them into 13 categories regular sketch entities, for which cannot be summarized in that category shall be classified as irregular sketch entities, and according to the types of sketch entities, the paper clarifies the sketch. The sketch consisting of the regular graphic element belongs to the first category, and others belong to the second category. The paper proposes one method based on the sketch entities and stroke structure, and the other way based on perceptual hashing technique, then realize two-stage identification algorithm of handwritten sketch. The paper has recognized the 150 kinds of sketch, of which there are 100 samples of each sketch, including 50 as the training sample,50 as the test sample, and the average recognition accuracy is 82.6%. The experimental results show that, the proposed algorithm has high recognition accuracy and efficiency.