Decomposition and recognition of playing volleyball action based on SVM algorithm
Yimao Yang · Journal of Interdisciplinary Mathematics · 2018
For the object-oriented high resolution playing volleyball action recognition, due to the large sample size and data, the recognition rate is not high, so the support vector machine (SVM) and its improvement are proposed. Firstly, carry out the principal component analysis (PCA) for the sample data of playing volleyball action to achieve reduced dimension; then, carry out the SVM classifier train for the sample data after reduced dimension, and obtain the optimum parameters of the reduced dimension data by the grid search method; finally, reset the SVM classifier parameter search range based on the original sample data so as to obtain the optimum SVM classifier parameters of the original sample data and realize the decomposition and recognition of playing volleyball action. The result shows that SVM algorithm can find the optimum SVM classifier parameters quickly and effectively, and obtain high accuracy of volleyball action recognition.