排球扣球动作的三维手势重建系统设计

为了精确、快速、低成本地获得运动员的排球扣球动作三维手势模型并对其进行数字化模拟,需要对排球扣球动作的三维手势进行重建系统设计。传统的基于多台Kinect的排球扣球动作三维手势重建系统设计方法,首先使用两台Kinect分别采集运动员排球扣球动作二维手势图像的点云数据,通过标定得到两台Kinect采集到的排球扣球动作二维手势图像姿态关系；然后基于标定结果,使用迭代最近点算法对两部分排球扣球动作三维手势图像的点云数据进行配准得到完整的排球扣球动作三维模型。存在重建成本高,噪声干扰严重导致重建系统失真严重的问题。为了降低噪声干扰,提高重建精确度,降低成本,提出一种基于立体视觉的排球扣球动作三维手势重建系统设计方法,首先通过传感器或摄像机等设备获取排球扣球动作二维手势图像,并对其进行平滑去噪；然后采用基于2D平面标靶的标定原理,建立摄像机成像的排球扣球动作三维手势图像几何模型并获得其参数；最后利用双目立体视觉算法实现对排球扣球动作三维手势图像的特征提取与匹配,完成三维手势重建。实验结果分析证明,所提方法可以提高排球扣球动作三维手势重建效率,提高重建精确度,降低噪声影响,具有更加广泛的使用价值。

Volleyball Spiking Action of 3d Gesture Reconstruction System Design

In order to accurate, fast, low cost to obtain the volleyball spiking 3d gesture of movement model and carries on the digital simulation, the need to 3d gesture of volleyball spiking motion reconstruction system design. Traditional based on more than one device of volleyball spiking action 3d gesture reconstruction system design method, the first to use two device respectively collect images of volleyball spiking two-dimensional gestures of movement of point cloud data, collected in two device is obtained by calibration of volleyball spiking posture movement two-dimensional gesture image; Then based on the results of calibration using iterative closest point algorithm for the two parts of volleyball spiking action three dimensional point cloud data of gesture image registration to get a complete 3d model of volleyball spiking action. High cost of reconstruction, noise is serious distortion of serious problems in reconstruction system. To reduce noise, improve the reconstruction accuracy, reduce costs, put forward a kind of volleyball spiking motion based on stereo vision 3d gesture reconstruction system design method, first by the movements of the sensors or cameras and other equipment for volleyball spiking two-dimensional gestures images, and carries on the smooth denoising; And then the calibration principle based on 2 d planar target, set up the camera imaging of volleyball spiking action 3d gesture images and obtain its geometric parameters; Finally using binocular stereo vision algorithm of volleyball spiking action 3d gesture image feature extraction and matching, complete 3d gesture to rebuild. Analysis of experimental results show that the proposed method can improve the efficiency of volleyball spiking action 3d gesture reconstruction, improve the reconstruction accuracy, reduce the noise influence, it has more extensive using value.