Recognition Method of Single Hand over Shoulder Shooting based on Multi-Dimensional Visual Images

Abstract

When the current action recognition methods recognize one hand over the shoulder shooting action, it will cause noise and holes, which will affect the recognition of one hand over the shoulder shot action, resulting in low accuracy rate of single hand over shoulder shot action recognition. Therefore, this paper proposes a recognition method of one hand over shoulder shot action based on multi-dimensional visual images. The image acquisition model of one hand over the shoulder shot is established, and the multi-dimensional visual image is used to mark the single hand over the shoulder shooting action; the image segmentation and feature extraction are preprocessed according to the marking results of the single hand shoulder shot action; the four-dimensional vector of the single hand over shoulder shot image is established to calculate the movement process of the one hand over the shoulder shot action and identify the single hand over shoulder shot action. The experimental results show that under the background of the same one handed shoulder shot and arm motion track, the arm motion track can be identified and the recognition accuracy rate can be detected. The identification method of single hand shoulder shot action can accurately identify the single hand shoulder shot action and the arm movement track, and the recognition accuracy rate is high.