基于骨架约束的人体轮廓图像变形算法  

Human contour image deformation algorithm based on skeleton constraint

作　　者：蔡雨轩 车文刚[1] CAI Yu-xuan;CHE Wen-gang(Faculty of Information Engineering and Automation,Kunming University of Science and Technology,Kunming 650500,China)

机构地区：[1]昆明理工大学信息工程与自动化学院,昆明650500

出　　处：《信息技术》2021年第4期69-74,共6页Information Technology

摘　　要：经典的基于控制点的移动最小二乘变形方法,没有考虑到图像变形过程中图像的拓扑结构,导致变形效果脱离现实人体轮廓结构,变得夸张且不符合常理。而骨架刚好能完整地保留图像轮廓的拓扑结构。提出一种在变形过程中加入骨架约束的变形算法,通过推导计算得到一个约束系数,使每一个像素点的移动相对于骨架存在一个约束,从而达到真实有效的变形。实验结果表明,文中方法可以使人体轮廓图像由瘦变胖的变形过程更加符合现实中人体曲线的变化情况。The control points based classic moving least squares deformation method does not take into account the topological structure of the image in the process of image deformation,causing the deformation effect to deviate from the contour structure of the actual human body,which becomes exaggerated and unreasonable.The skeleton could just keeps the complete topological structure of the image contour.This paper proposes a deformation algorithm that adds the skeleton constraint in the deformation process,and obtains a constraint coefficient through derivation and calculation,which allows the movement of each point to have a constraint relative to the skeleton,so as to achieve an effective deformation.The experimental results show that the method can make the process from thin to fat of human body contour more in line with the change of the human body curve in reality.

关 键 词：计算机技术 图像变形 骨架约束 移动最小二乘法 人体轮廓图像 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]