Implementation of Instant Numeral Recognition in Space Trajectory

This study implemented the hand gesture digit recognition function by using the Kinect motion sensor. This interactive motion-sensing technology was employed to track, record, and convert the trajectories of handwritten numerals in space into corresponding shapes of digits on the plane. Subsequently, the feature matching method was used to achieve digit recognition. In the recognition process, first binarization and Hilditch thinning algorithm were performed on the region for recognition, and the segmentation algorithm proposed by G.E.M.D.C. Bandara etc. was used to perform segmentation of the region after thinning. This paper proposes the improvement of the conditions of the arc type of the discriminant to achieve feature extraction and enhance the digit recognition rate.