A Study on Posture Correction Based on Computer Vision

Abstract

During basketball sport, the movement of a joint is quite complex. And if move fast, it is difficult to constrain by fix algorithm the subtle angle changes between joints. Traditional sports vision modeling method is unable to describe the moving changes of the small areas which causes the unsatisfactory measuring effect of subtle posture in motion. This paper proposes a measurement method for three-dimensional motion posture of basketball athletes. It converts the constrained optimization problem for motion parameters as a nonlinear minimization problem by optimizing human motion parameters; uses L-M motion constraints parameter to provide fast convergent regularization method, in order to seek motion and structural parameters matrix of non-rigid basketball sport and complete three-dimensional measurement for motion parameters. The simulation results show that the method can accurately measure the 3D movement parameters of the athletes.