A Contour Extraction Method for Teaching Aerobics Dance Movements Based on Posture Data Sequence

The changes in fitness posture have a significant impact on the construction of human motion models, which may reduce the accuracy of joint feature recognition and make it difficult to effectively extract the edge contours of motion. Therefore, a contour extraction method for teaching aerobics dance movements based on posture data sequences has been proposed. Collect teaching images of fitness, bodybuilding, and dance activities and preprocess them; Based on posture data sequences, a human motion model is constructed using sparse MEMS human motion posture and position estimation techniques; By iteratively determining the positions of skeletal joints in fitness exercises through CNN, filling in missing nodes and using PCA algorithm for dimensionality reduction, the image contour extraction results are matched with the human motion model to achieve high-precision extraction of fitness dance teaching contours. The experimental results show that the integrity extracted by our method is consistently above 94%. Contour extraction has high stability, obvious advantages, and certain application value.