武术难点动作图像目标实时提取仿真  

Martial Arts Difficult Motion Image Target Real Time Extraction Simulation

作　　者：艾泽秀[1] 

机构地区：[1]成都体育学院武术系,四川成都610041

出　　处：《计算机仿真》2017年第8期289-292,共4页Computer Simulation

摘　　要：对武术难点动作图像目标的实时提取,能够有效判断武术动作的完成质量。对武术难点动作的图像进行实时提取,需要提取动作轮廓特征,与数据库中的轮廓进行匹配。传统方法根据动作图像像素与周围像素之间的关系,提取人体武术难点动作图像图像的边缘特征,但没有将其与数据库中的轮廓进行匹配,导致提取结果误差大。提出基于融合深度图的人体高难度武术难点动作图像实时提取方法。利用Kinect方法得到动作RGB彩色和深度图像,并对图像轮廓图像进行平滑处理,融合于细化方法提取图像轮廓骨架,选取1组深度图像序列,提取动作图像轮廓特征后与轮廓数据库中的轮廓进行匹配,得到相应的骨架,将骨架作为动作图像实时提取结果。仿真结果表明,所提方法实时提取精确度高,为广大教练员和运动员掌握人体高难度武术难点动作图像技术提供了有力的依据。A real - time extraction method for difficult motion image of martial art with high difficulty is proposed based on integrating with depth map. Firstly, the Kineet method is used to obtain color image and depth image of mo- tion RGB and smoothed image contour. The skeleton of image contour integrated with refining method is extracted to select a group of sequence of depth image. Then, the contour feature of motion image is extracted to match contour in contour database and the corresponding skeleton is obtained. Finally,the skeleton is used as real -time extraction re- suits of motion image. Simulation results show that the method has high extraction precision. It provides powerful gist for coaches and athletes mastering image technology of difficult motion in high difficulty martial art.

关 键 词：武术难点动作 图像目标 实时提取 

分 类 号：G842.19[文化科学—体育训练]