An Efficient Motion Registration Method Based on Self-Coordination and Self-Referential Normalization

• DOI: 10.3390/electronics11193051
• Publication Date: 2022-09-26T03:13:27Z
• Authors: Yuhao Ren; Bochao Zhang; Jing Chen; Liquan Guo; Jiping Wang
• Keywords: Normalization; Position (finance)

Action quality assessment (AQA) is an important problem in computer vision applications. During human AQA, differences body size or changes position relative to the sensor may cause unwanted effects. We propose a motion registration method based on self-coordination (SC) and self-referential normalization (SRN). By establishing coordinate system using part of as normalized reference standard process raw data, standardization distinguishability data are improved. To demonstrate effectiveness our method, we conducted experiments KTH datasets. The experimental results show that improved classification accuracy KNN-DTW network for KTH-5 from 82.46% 87.72% KTH-4 89.47% 94.74%, it tsai-MiniRocket 91.29% 93.86% 94.74% 97.90%. can reduce above effects improve action network. This study provides new idea improving AQA-related algorithms.