摘要
人体测量技术是服装制造业中一个重要的环节,一种便捷、高效率、高精度的人体测量技术可为服装行业的制造提供数据基础。针对传统二维图像人体测量方法中过程繁琐、耗时等问题,提出将基于TensorFlowj.s的BodyPix应用到人体测量中的新思路,在BodyPix分割工具获取到人体二维分割图像的基础上通过像素点扫描的方法把人体各部位的图像尺寸转换为真实尺寸,可以有效提高服装行业中在面对批量服装定制时量体的速度。实验结果中人体的围度误差在-0.969^+0.855,长度误差在-0.944^+0.723。有效提高了人体测量的精准性和测量速度,表明了将BodyPix应用到人体测量中的鲁棒性。
Anthropometry technology is an important link in clothing manufacturing industry.A convenient,efficient and high-precision anthropometry technology can provide data basis for the manufacturing of clothing industry.Based on the traditional two-dimensional image anthropometric method in problem such as tedious,time-consuming process,put forward will be based on TensorFlow.Js BodyPix is applied to the new way of thinking,in body measurements in BodyPix segmentation tools get to the human body on the basis of two-dimensional image segmentation by pixel scanning method converts the image size of each part of the human body as the real size,can effectively increase the volume in the face of the bulk garment customization in apparel industry body.In the experimental results,the range error of human body was-0.969~ +0.855,and the length error was-0.944~ +0.723.The accuracy and speed of anthropometric measurement are improved,which indicates the robustness of BodyPix applied to anthropometric measurement.
作者
李科
毋涛
Li Ke;Wu Tao(College of Computer Science,Xi'an Polytechnic University,Xi′an 710048,China)
出处
《国外电子测量技术》
2020年第7期89-93,共5页
Foreign Electronic Measurement Technology