Abstract
Football is a sport that requires sportsmen to have both physical strength and physical features. It must consider the distinctions between individuals and then provide targeted training. Football players can perform better on the field with targeted scientific training, but scientific training is based on identifying football players' technical actions and behaviors. Deep learning allows machines to emulate the behavior of humans, like sight, hearing, and thought. It solves a wide range of complicated pattern recognition issues. The deep learning procedure, in particular, is distinctive in its capacity to recognize images with great precision and offers technical assistance for analyzing and recognizing football players' behavior actions. However, traditional football action recognition mainly uses the standard local binary pattern (LBP) for recognition. In image recognition, problems include the high dimension of football technical action recognition data and inaccurate recognition. Principal component analysis (PCA) can be used to perform dimensionality reduction analysis on the technical action behavior of football players to reduce the amount of calculation in the process of technical action recognition. This paper compared and analyzed football players' technical action behavior recognition based on the PCA-LBP algorithm and the traditional LBP recognition. The data comparing the two algorithms are based on data from 200 football players at a football match in 2020. This paper mainly counts the specific stadium information of football players and the data samples of football technical action recognition. In addition, it uses the four technical actions of kicking, dribbling, stopping, and fake action as indicators to evaluate the accuracy of technical action recognition. The experimental results showed that the recognition accuracy of the PCA-LBP algorithm is 2% higher than that of the LBP algorithm when the number of kicking action recognition is 50 times. When the number of recognition times was 300, the recognition accuracy of the PCA-LBP algorithm was 24% higher than that of the LBP algorithm. The PCA-LBP algorithm also has higher recognition accuracy when comparing dribbling, stopping, and fake action. Therefore, using PCA to decrease the dimension of the LBP algorithm can enhance the accuracy of the recognition of the technical action behavior of football players.