乒乓球运动员发球姿势三维图像校正仿真

发布时间：2018-06-27 02:26

本文选题：乒乓球 + 发球姿势　；参考：《计算机仿真》2017年11期

【摘要】：对乒乓球运动员发球姿势三维图像的校正,能够有效提升乒乓球比赛质量。对三维图像的校正,需要将图像分解为基本分量与细节分量,对分离出的三维图像细节分量时域曲线进行梯度加权均值滤波,完成三维图像的校正。传统方法先设计理想基线和光心,估计乒乓球发球姿势三维图像细节深度信息,但忽略了对图像细节进行梯度加权均值滤波,导致校正精度偏低。提出基于三边滤波的乒乓球发球姿势三维图像校正方法。构造乒乓球发球姿势结构保持的三维图像运动伪影校正模型,该模型由图像非局部均值正则项和块相似保真项构成,再将乒乓球发球姿势三维图像分解为基本分量与细节分量,对三维图像细节分量的时域曲线进行梯度加权均值滤波。实验结果表明,所提方法能够较好地保留图像细节特征,且各项图像客观评价参数具有大幅度提升。
[Abstract]:The correction of three-dimensional images of serving position of table tennis players can effectively improve the quality of table tennis matches. It is necessary to decompose the image into basic component and detail component to correct the 3D image. The gradient weighted mean filter is applied to the time domain curve of the detail component of the separated 3D image to complete the correction of the 3D image. The traditional method first designs the ideal baseline and the center of light to estimate the depth information of the three-dimensional image of table tennis serve posture, but neglects the gradient weighted mean filtering of the image details, which leads to the low accuracy of correction. A three-dimensional image correction method for table tennis serving posture based on trilateral filtering is proposed. The motion artifact correction model of table tennis serving posture structure is constructed. The model is composed of non-local mean canonical items and block similar fidelity terms, and then decomposes the three-dimensional images of table tennis serve posture into basic components and detail components. The gradient weighted mean filter is applied to the time domain curve of the detail component of the 3D image. The experimental results show that the proposed method can preserve the detailed features of the image well, and the objective evaluation parameters of each image can be greatly improved.
【作者单位】：南京晓庄学院体育学院;
【分类号】：G846;TP391.41

【相似文献】