基于机器视觉的健美操跳跃动作特征提取方法研究

Research onFeature Extraction Method of Aerobics Jump Action Based on Machine Vision

由于传统健美操跳跃动作特征提取方法存在动作方位角度识别准确率低、特征提取率低和特征提取效率低的问题,提出基于机器视觉的健美操跳跃动作特征提取方法。通过机器视觉获取健美操视频,提取健美操视频的熵值序列和音乐特征,融合以上特征提取健美操动作关键帧,利用高斯混合模型对关键帧进行处理,消除健美操关键帧的背景;采用阈值识别算法识别健美操跳跃动作,结合Harris3D算子建立健美操跳跃动作序列势函数,在此基础上,利用AdaBoost算法提取健美操跳跃动作特征。实验结果表明,所提方法的动作方位角度识别准确率高、特征提取率高、提取效率高。

Because the traditional method of aerobics jumping action feature extraction has the problems of low recognition accuracy,low feature extraction rate and low feature extraction efficiency,a feature extraction method of aerobics jumping action based on machine vision is proposed.It obtains aerobics videos through machine vision,extracts the entropy sequence and music characteristics of aerobics videos,combines the above features to extract key frames of aerobics,uses Gaussian mixture model to process the key frames,eliminates the background of aerobics key frames;adopts threshold.The recognition algorithm recognizes the aerobics jumping action,and combines the Harris3D operator to establish the aerobics jumping action sequence potential function.On this basis,the AdaBoost algorithm is used to extract the characteristics of the aerobics jumping action.The experimental results show that the proposed method has high recognition accuracy,high feature extraction rate and high extraction efficiency.