基于目标区域约束的人体手臂运动轨迹图像分割
|
摘要
针对传统分割方法存在分割完整性不足、分割耗时较长以及分割精度较差的问题,提出基于目标区域约束的人体手臂运动轨迹图像分割方法.采取多阈值分割方法,对目标区域约束阈值进行设定,根据阈值取值,得出图像阈值分割曲面图.通过对比巴氏距离系数描述轨迹目标区域与候选目标区域之间的相似度,根据Mean Shift检索邻域范围内密度评估的极大值,迭代上述过程,不断更新运动轨迹目标,根据图像轨迹目标跟踪结果,利用目标区域约束法实现图像分割.结果表明,利用该方法后图像分割区域较为完整,耗时较短,分割精度明显提高.
Aiming at the problems of insufficient segmentation integrity,long time consuming and poor segmentation precision existing in traditional segmentation methods,an image segmentation method for the motion trajectory of human arm based on target region constraint was proposed. With the multi-threshold segmentation method,the constraint threshold of target region was set. According to the threshold value,the image threshold segmentation surface diagram was obtained. By comparing the Pasteurian distance coefficient,the similarity between the locus target region and the candidate target region was described.According to Mean Shift,the maximum value of density evaluation in neighboring regions was retrieved,the abovementioned process was iterated and the motion trajectory target was continuously updated.According to the tracking results of image trajectory target,the image segmentation was implemented with the target region constraint method. The results showthat the image segmentation region is relatively complete with the as-proposed method,consuming less time with apparently improved accuracy for image segmentation.
Aiming at the problems of insufficient segmentation integrity,long time consuming and poor segmentation precision existing in traditional segmentation methods,an image segmentation method for the motion trajectory of human arm based on target region constraint was proposed. With the multi-threshold segmentation method,the constraint threshold of target region was set. According to the threshold value,the image threshold segmentation surface diagram was obtained. By comparing the Pasteurian distance coefficient,the similarity between the locus target region and the candidate target region was described.According to Mean Shift,the maximum value of density evaluation in neighboring regions was retrieved,the abovementioned process was iterated and the motion trajectory target was continuously updated.According to the tracking results of image trajectory target,the image segmentation was implemented with the target region constraint method. The results showthat the image segmentation region is relatively complete with the as-proposed method,consuming less time with apparently improved accuracy for image segmentation.
引文
[1]张嫣.基于视觉分析的篮球投篮动作标准化判断方法研究[J].现代电子技术,2017,40(3):55-58.(ZHANG Yan. Research on standardized judgement method of basketball shooting action based on visual analysis[J]. M odern Electronic Technology,2017,40(3):55-58.)
[2]杨信廷,孙文娟,李明,等.基于K均值聚类和开闭交替滤波的黄瓜叶片水滴荧光图像分割[J].农业工程学报,2016,32(17):136-143.(YANG Xin-ting,SUN Wen-juan,LI ming,et al. Water droplet fluorescence image segmentation of cucumber leaves based on K-means clustering and open/close alternating filter[J]. Journal of Agricultural Engineering,2016,32(17):136-143.)
[3]袁小翠,吴禄慎,陈华伟.基于Otsu方法的钢轨图像分割[J].光学精密工程,2016,24(7):1772-1781.(YUAN Xiao-cui,WU Lu-shen,CHEN Hua-wei. Rail image segmentation based on Otsu method[J]. Optical Precision Engineering,2016,24(7):1772-1781.)
[4]许冰,牛燕雄,吕建明.复杂动态场景下目标检测与分割算法[J].北京航空航天大学学报,2016,42(2):310-317.(XU Bing,NIU Yan-xiong,LJian-ming. Target detection and segmentation algorithm in complex dynamic scene[J]. Journal of Beijing University of Aeronautics and Astronautics,2016,42(2):310-317.)
[5]任侃,王伟杰,钱惟贤,等.移动相机下的地面运动目标分割技术[J].北京邮电大学学报,2017,40(1):46-49.(REN Kan,WANG Wei-jie,QIAN Wei-xian,et al.Ground moving object segmentation technique under moving camera[J]. Journal of Beijing University of Posts and Telecommunications,2017,40(1):46-49.)
[6]宋琳,高满屯,王三民,等. CV-GAC模型与图割优化的运动目标检测和分割[J].机械科学与技术,2017,36(1):102-107.(SONG Lin,GAO Man-tun,WANG San-min,et al.CV-GAC model and graph cut optimization for moving object detection and segmentation[J]. Mechanical Science and Technology,2017,36(1):102-107.)
[7]王科飞.基于模糊聚类算法的体育运动视频图像分析应用[J].现代电子技术,2017,40(9):39-42.(WANG Ke-fei. Application of sports video image analysis based on fuzzy clustering algorithm[J].M odern Electronic Technology,2017,40(9):39-42.)
[8]王瑜,闫沫.基于Wasserstein距离和分裂Bregman方法的图像分割算法[J].电子设计工程,2017,25(2):140-144.(WANG Yu,YAN Mo. An image segmentation algorithm based on Wasserstein distance and split Bregman method[J]. Electronic Design Engineering,2017,25(2):140-144.)
[9]徐天芝,张贵仓,贾园.基于形态学梯度的分水岭彩色图像分割[J].计算机工程与应用,2016,52(11):200-203.(XU Tian-zhi,ZHANG Gui-cang,JIA Yuan. Watershed color image segmentation based on morphological gradient[J]. Computer Engineering and Applications,2016,52(11):200-203.)
[10]乔良.三维图像视觉下运动关键特征的提取方法改进[J].计算机仿真,2017,34(1):216-219.(QIAO Liang. An improved method for extracting key features of motion in 3D image vision[J]. Computer Simulation,2017,34(1):216-219.)
[11]杨品林.彩色图像数据库中目标特征数据挖掘方法[J].沈阳工业大学学报,2018,40(1):60-64.(YANG Pin-lin. Method of object feature data mining in color image database[J]. Journal of Shenyang University of Technology,2018,40(1):60-64.)
[2]杨信廷,孙文娟,李明,等.基于K均值聚类和开闭交替滤波的黄瓜叶片水滴荧光图像分割[J].农业工程学报,2016,32(17):136-143.(YANG Xin-ting,SUN Wen-juan,LI ming,et al. Water droplet fluorescence image segmentation of cucumber leaves based on K-means clustering and open/close alternating filter[J]. Journal of Agricultural Engineering,2016,32(17):136-143.)
[3]袁小翠,吴禄慎,陈华伟.基于Otsu方法的钢轨图像分割[J].光学精密工程,2016,24(7):1772-1781.(YUAN Xiao-cui,WU Lu-shen,CHEN Hua-wei. Rail image segmentation based on Otsu method[J]. Optical Precision Engineering,2016,24(7):1772-1781.)
[4]许冰,牛燕雄,吕建明.复杂动态场景下目标检测与分割算法[J].北京航空航天大学学报,2016,42(2):310-317.(XU Bing,NIU Yan-xiong,LJian-ming. Target detection and segmentation algorithm in complex dynamic scene[J]. Journal of Beijing University of Aeronautics and Astronautics,2016,42(2):310-317.)
[5]任侃,王伟杰,钱惟贤,等.移动相机下的地面运动目标分割技术[J].北京邮电大学学报,2017,40(1):46-49.(REN Kan,WANG Wei-jie,QIAN Wei-xian,et al.Ground moving object segmentation technique under moving camera[J]. Journal of Beijing University of Posts and Telecommunications,2017,40(1):46-49.)
[6]宋琳,高满屯,王三民,等. CV-GAC模型与图割优化的运动目标检测和分割[J].机械科学与技术,2017,36(1):102-107.(SONG Lin,GAO Man-tun,WANG San-min,et al.CV-GAC model and graph cut optimization for moving object detection and segmentation[J]. Mechanical Science and Technology,2017,36(1):102-107.)
[7]王科飞.基于模糊聚类算法的体育运动视频图像分析应用[J].现代电子技术,2017,40(9):39-42.(WANG Ke-fei. Application of sports video image analysis based on fuzzy clustering algorithm[J].M odern Electronic Technology,2017,40(9):39-42.)
[8]王瑜,闫沫.基于Wasserstein距离和分裂Bregman方法的图像分割算法[J].电子设计工程,2017,25(2):140-144.(WANG Yu,YAN Mo. An image segmentation algorithm based on Wasserstein distance and split Bregman method[J]. Electronic Design Engineering,2017,25(2):140-144.)
[9]徐天芝,张贵仓,贾园.基于形态学梯度的分水岭彩色图像分割[J].计算机工程与应用,2016,52(11):200-203.(XU Tian-zhi,ZHANG Gui-cang,JIA Yuan. Watershed color image segmentation based on morphological gradient[J]. Computer Engineering and Applications,2016,52(11):200-203.)
[10]乔良.三维图像视觉下运动关键特征的提取方法改进[J].计算机仿真,2017,34(1):216-219.(QIAO Liang. An improved method for extracting key features of motion in 3D image vision[J]. Computer Simulation,2017,34(1):216-219.)
[11]杨品林.彩色图像数据库中目标特征数据挖掘方法[J].沈阳工业大学学报,2018,40(1):60-64.(YANG Pin-lin. Method of object feature data mining in color image database[J]. Journal of Shenyang University of Technology,2018,40(1):60-64.)