| 基于双目视觉的运动位姿误差检测方法研究 |
| |
| 引用本文： | 邰玉明,陶存然.基于双目视觉的运动位姿误差检测方法研究[J].宜春学院学报,2021(3):100-103. |
| |
| 作者姓名： | 邰玉明 陶存然 |
| |
| 作者单位： | 滁州学院体育学院 |
| |
| 基金项目： | 国家级大学生创新训练项目“体育中考分值变化对初等体育教育的影响研究”(项目编号:2018CXXL097);滁州学院教学团队项目“‘筑梦’创新创业指导团”(项目编号:2016jxtd04)。 |
| |
| 摘 要： | 为了提高人体运动位姿误差检测能力,提出基于双目视觉的运动位姿误差检测方法。采用双目视觉跟踪融合识别方法对运动位姿的样本动态特征点进行采样,以全局人体姿态信息为候选样本,进行运动位姿双目视觉特征高分辨提取,采用模板匹配方法,构建人体部位姿态候选样本轮廓分布集,获取模板大小,根据模板大小计算运动位姿的误差概率分布,利用相邻图像帧之间的运动特征分布集,构建运动位姿图像的位置信息检测模型,在此基础上,采用自适应颜色覆盖方法对运动位姿的误差概率分布做极小化处理,完成对人体运动位姿误差检测。仿真结果表明,采用该方法进行运动位姿误差检测的特征分辨能力很好,降低了检测误差,提高了运动位姿双目视觉跟踪识别能力。
|
| 关 键 词： | 双目视觉 运动 位姿 误差 检测 |
Research on the Motion Pose Error Detection Method Based on Binocular Vision |
| |
| TAI Yu-ming,TAO Cun-ran.Research on the Motion Pose Error Detection Method Based on Binocular Vision[J].Journal of Yichun University,2021(3):100-103. |
| |
| Authors: | TAI Yu-ming TAO Cun-ran |
| |
| Institution: | (College of Physical Education,Chuzhou University,Chuzhou 239012,China) |
| |
| Abstract: | In order to improve the detection ability of human motion pose error,a method based on binocular vision is proposed.Using binocular vision tracking fusion recognition method to sample the dynamic feature points of the moving pose samples,taking the global human posture information as the candidate samples,extracting the high-resolution binocular visual features of the moving pose.Using template matching method,constructing the contour distribution set of the human body posture candidate samples,obtaining the template size,and calculating the error probability of the moving pose according to the template size Distribution,using the distribution set of motion features between adjacent image frames,the position information detection model of motion pose image is constructed.On this basis,the adaptive color coverage method is used to minimize the error probability distribution of motion pose and complete the detection of human motion pose error.The simulation results show that the feature resolution ability of the method is very good,the detection error is reduced,and the binocular vision tracking and recognition ability of the motion pose is improved. |
| |
| Keywords: | binocular vision motion pose error detection |
| 本文献已被 维普 等数据库收录！ |
|
