摘要
传统的人体腰围测量方法存在精度低和成本高等问题。为解决该问题,本文提出基于特制腰带的人体腰围测量方法。该方法首先制作一条由不同颜色的等距方块构成的软磁条腰带,接着,利用Harris法和FAST方法对角点进行检测和筛选,并利用二次曲线拟合进一步搜索角点,再利用Canny算子提取图像边缘轮廓并判断边缘的闭合性,通过不闭合的边缘寻找腰带的截止线,并利用BP神经网络进行方块颜色识别,最后,按基础腰围、叠加腰围和附加腰围三部分相加得到人体腰围。实验结果表明,利用本文方法测得的腰围绝对误差在-0.70~0.70 cm之间,满足服装业生产中±1.5 cm的要求。
Traditional waist circumference measurement methods have the problems of low accuracy and high cost.To tackle this issue,we propose waist circumference measurement method based on the custom waistband.Firstly,we make the soft magnetic waistband with different color and equidistant blocks.Next,we detect and select the corners with Harris and FAST algorithms and further search the corners with the fitting quadratic curve.Then,we extract the edge contour of the image to judge the edge closure with Canny operator and find the cutoff line in the waistband through the unclosed edge,and recognize the color of blocks with BP network.Finally,we acquire the waist circumference with three parts including basic waist circumference,superposed waist circumference and additional waist circumference.The experimental results show that the absolute error of waistline measured by the proposed method is between-0.70~0.70 cm,which meets the requirements of the±1.5 cm in the garment industry.
作者
李咏豪
马燕
LI Yonghao;MA Yan(College ofCompuer Science and Engineering,Nanjing University of Science&Technology,Nanjing 210094,China;College of Information,Mechanical and Electrical Engineering,Shanghai Normal University,Shanghai 200234,China)
出处
《电视技术》
2019年第8期15-18,21,共5页
Video Engineering
基金
国家自然科学基金资助项目(61373004)
