Sphere motion mode identification method and device, and storage medium

Abstract

The present invention discloses a sphere motion mode identification method and device, and a storage medium. The method comprises the steps of: employing local multi-frame track information, and defining different position and displacement convolution kernels to search the best matching position and displacement mode; and determining and obtaining a table sphere motion mode according to the best matching position and displacement mode. The method employs the local multi-frame track information to remove the noise interference in a track and stably identify a motion state of a sphere; and moreover, the method does not need to establish positive and negative sample database to train a model, and is simple to operate. The method is wide in application range, and is suitable for identificationof lots o sphere motion modes.

Description

technical field [0001] The present invention relates to the technical field of smart courts, in particular to a method, device and storage medium for ball movement pattern recognition. Background technique [0002] The recognition of the motion pattern of the ball is an indispensable part of the intelligent stadium, and it is the basis for the analysis of the ball motion trajectory, which plays a very important role. At present, the following two schemes are usually adopted for the recognition of the motion pattern of the sphere: [0003] One is to establish a sample library and train the hidden Markov model to identify the movement pattern of the sphere; the other is to establish a sample library and train the corresponding neural network classifier to identify the movement pattern of the sphere. [0004] However, existing methods such as hidden Mark Lev models and neural network classifiers need to establish a positive and negative sample library for training, so the proc...

AI Technical Summary

Problems solved by technology

[0004] However, existing methods such as hidden Mark Lev models and neural network classifiers need to establish a positive and negative sample library for training, so the process is more complicated

In addition, the existing part of the sphere trajectory analysis method uses the trajectory information of two consecutive frames to identify the sphere motion pattern, which is easily disturbed by noise, resulting in unstable recognition

Images