Research Of Human Body Anthropometric Measurement Based On Image

With the rapid development of electronic commerce,the measurement of human body size has gradually changed from manual measurement to non-contact anthropometric measurement.As a non-contact two-dimensional anthropometric technology,image-based measurement of human body size has the advantages of fast,simple and low cost,and has higher practical value for the garment industry.It has gradually become the research trend of non-contact anthropometric technology.This paper studies the measurement of human body size based on image,introduces the whole process of measurement of human body size from three aspects: segmentation of human image,extraction of human key points and calculation of human body size.The main research contents are as follows:(1)Crotch detection based on frontal body image: firstly,analyze the specific location of the crotch point in the front image of the human body,and establish two geometric models of the front human body image: the concentric circle model of crotch and the rectangular symmetrical model of crotch.The crotch point in the front human body image is detected according to the geometric model of human crotch.Finally the crotch point is detected accurately from the front human body image.Experiments have shown that the method can accurately extract the crotch point in the front human body image.(2)The human body segmentation method based on the superpixel and the proportion of human body: The human body image is segmented into several superpixel blocks by using the superpixel segmentation method.By analyzing the proportion rule of human body,get the human proportion relation between head and each part and delimit the approximate range of human body.Within the scope of the calibrated human body,the superpixel block is marked as the background block and the foreground block,and the background block is constantly merged through the iterative method,finally the segmentation result of the front and side human body image is obtained.The experimental results show that the proposed method is effective for the segmentation of front and side human body image when the background complexity changes.(3)Feature point detection method based on human body proportion: firstly,the human body edge contour is extracted and coded by image segmentation algorithm.Find all the turning point on the human body edge contour,the human body contour within the height range of neck,shoulder,chest,waist and hip was screened for turning points according to the proportion of human body and feature point condition determination,the turning points that conform to the definition of human feature points are marked as candidate human key points.Then,according to the detailed definition of each characteristic part in the measurement method of human body,the feature points of each characteristic part are further calibrated accurately.Finally,the feature points of the front and side human body edge were obtained.(4)Body circumference fitting calculation: firstly,the horizontal and vertical dimension data of human body are obtained by using the feature points of human body detected by human edge;a linear regression model was used to calculate shoulder width,a binary linear regression model was used to calculate neck circumference,and a logarithmic model was used to calculate chest circumference,waist circumference and hip circumference;finally,the body circumference of each characteristic part is calculated.The error analysis shows that the size data of human body calculated by the fitting model can satisfy the requirements of anthropometric measurement.