Real-time Feedback and Evaluation Algorithm for Children's Digital Writing Practice

Abstract

At present, the application of handwritten numeral practice is common. However, these applications only focus on the imitation practice of drawing red, that is, they only focus on the similarity of shape, and they don't pay attention to the basic stroke order and stroke number. What's more, they can't judge the quality of digital writing and give real-time feedback. In this paper, the handwriting of children is preprocessed by mathematical morphology operation, and then the convolution neural network is used to recognize the number. After recognition, the handwriting is processed with fine lines, and the improved Hilditch algorithm is used for skeleton extraction. The next the features of handwritten numerals are extracted, such as corners, lines, arcs, the proportion and position of handwriting pixels. These features are used in the fuzzy comprehensive evaluation method to realize the real-time evaluation and feedback of handwritten numbers, and truly achieve the purpose of digital writing practice. Experiments show that the algorithm has good real-time and accuracy, and can improve the efficiency and enthusiasm of children's independent practice of digital writing.