小差异运动图像中关键兴趣姿态点的差异特征分离  

Separation of Characteristics for Key Interest Profile Point in Small Differences Motion Image

作　　者：刘旭东[1] 

机构地区：[1]许昌学院体育学院,河南许昌461000

出　　处：《科技通报》2014年第2期139-142,共4页Bulletin of Science and Technology

摘　　要：在一些连续帧的运动图像中,连续图像帧之间运动姿态点差异很小,形成像素粘连。这些细微的粘连差异特征不能形成很好的特征分离,将影响后期的处理精度。提出一种小差异运动图像中关键兴趣姿态点的差异特征分离方法,通过计算连续运动图像帧的轮廓像素空间位置参数,设置轮廓像素方向分量,计算任意像素两端的灰度值,通过运动图像窗口的移动,进行像素灰度累加,完成连续帧图像轮廓区域空间位置参数,结合连续图像间的差异拟合参数,破处粘连,实现小差异运动图像中关键兴趣姿态点的差异特征分离。仿真实验结果表明,该方法可以对高精度连续图像帧之间的运动特征差异进行分离,提高特征提取精度。In some continuous motion images, motion point difference between consecutive image frames is small, pixel adhesions. The adhesion of the subtle differences in features could not form a good separation, will affect the processing precision of late. Put forward a small difference in the image motion posture point discrepancy in key characteristics of separation methods, through the calculation of contour pixels of image frames continuous movement space position param-eters, set up the contour pixel component, calculated at the ends of the arbitrary pixel gray value, through a moving im-age of the window, pixel gray accumulation, complete the continuous image contour area spatial location parameters, com-bining the difference between consecutive image fitting parameters, virginity adhesion, implement key interest profile in small differences in motion image points difference features of separation. The simulation results show that the method can be of high precision movement characteristics of the difference between consecutive image frames, improve the fea-ture extraction accuracy.

关 键 词：小差异 运动图像 特征分离 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]