Human pose estimation is one of the hot topics in the field of computer vision,and can be used for pedestrian detection,human activity analysis,human-computer interaction and video surveillance and so on. It is difficult to robustly estimate the human pose under the complex background for existing estimation methods of human pose,which is partially due to the lack of good features to separate the foreground human from the complex background. Aiming at the deficiencies mentioned above,a pose segmentation and estimation method is presented. The human is segmented from the background by semantic segmentation. Then shape context method is adopted to obtain the optimal human pose in the pictorial structure. Experimental results show that the proposed method can get the pose estimation,overcome the interference from background,and obtain a better body estimation than the existing method under complex backgrounds. So it can be success to estimate the body pose from the image in a complex background.%计算机视觉领域中现在有一个非常热门的问题就是人体的姿态估计，它可用于行人检测、人体活动分析、人机交互以及视频监控等方面。目前对于图像的人体姿态的估计方法在处理较复杂的背景的时候难以得到理想的效果，其原因在于这些方法不好区分人体和复杂背景，从而无法得到其想要的特征值供其使用。针对这一不足，提出一种姿态分割估计方法。该方法将人体分割后去除复杂背景的影响，并且在图结构模型中，结合使用形状上下文特征的方法进行训练对比，求解得出最优的人体姿态。实验结果表明，该方法可以较好地在复杂背景下获得人体的姿态估计，更好地克服背景带来的干扰，得到较现有方法更加理想的人体估计结果，从而把人体的姿态从复杂的背景图像中给成功地估计出来。
展开▼
