A Quantitative Assessment Study of Dynamic Symmetry and Aesthetic Value in Dance Theater Performance

Abstract

- In order to enhance the aesthetic value of dance drama performance, this paper is based on the human gesture recognition technology, extracting angle and gesture distance features, and carrying out normalization to accurately mark the human gesture of dance drama performance. A directed graph model in chronological order is constructed, and time is integrated into the model as a dynamic element. Describe the connection relationship between space and time in the sequence of human skeleton through the spatio-temporal graph, and establish feature vectors containing coordinate information and confidence level to get dynamic spatio-temporal symmetry. Select the nearest elements in the Top codebook to quantize the coding process, and finally get the quantization result of aesthetic value. Comparison with the three methods of convolutional neural network, optical flow method and dynamic temporal regularization reveals that the similarity mean value of this paper's method is 0.56, and the quantization process is more stable. During the iterative process of dynamic symmetry calculation, a high recognition rate of 0.99 is achieved, and the recognition accuracy can reach up to 100%. In addition the results of assessing the symmetry of movements in dance drama performance such as overall aesthetics have a smaller gap with the expert ratings, further proving the effectiveness of the constructed quantitative assessment model.