Design And Implement Of A 3D Human Body Measurement System Based On Multiple Kinects

With the development of the society and economy,the demand for clothing is becoming higher and higher.The customization of clothing and the individualization of the design have also become a trend.In years,the 3D non-contact measurement technology have developed rapidly,design and cutting the dress according to one's figure becoming more and more intelligent.Nowadays,the mainstream non-contact three-dimensional human body measurement equipment has high precision but high price,and it is difficult to operate,most of them also need large area and harsh environment,thus,they are not well popularize among the people and the ordinary business.To solve the above problems,we designed and implemented a non-contact 3D body measurement system based on multiple Microsoft's Kinect v2 devices,which has the characteristics of short time consumption,less space,low cost and high accuracy.This paper makes a thorough study of the following points:(1)An image enhancement algorithm was added to the calibration of the camera to improve the corner recognition rate in the dual target setting process(2)Depth data processing.We analyzed the current depth image process algorithm,and we improved the median blur algorithm which can smooth the depth images and maintain the edges.(3)Improved the ICP algorithm.To get an accurate model and decrease the calculate time,we downsample the point cloud data and smooth them,then fusion the point cloud data roughly,after the ICP algorithm,refine the results.At last,we have compared the results of the machine measurement and the actual measurement,and verify the effectiveness and error of the system,through the results,the system has higher accuracy and less scan time compared to little one,and compared with the trend equipment,the price is lower,and it is less affected by the environment.It has a broad prospect in practical application.