| With the development of electronic and image processing technology,study and application of non-contact body measurement technology become mature gradually.3D and 2D non-contact body measurement system have been developed and applied in practice.The 3D non-contact body measurement system has the advantages of extracting more data,fast calculating speed,etc.However,its high equipment cost limits the promotion and application of this technology.Because of its low euipment cost,high measurement speed,high efficiency and easy to grasp measurement techniques,the 2D non-contact body measurement system will be widely used in garment enterprises,especially in the small and medium sized garment enterprises.The results of body size measurement must be consistent with the requirements of garment structure design.However,for the current non-contact body measurement system,the problem that the measurement results of some characteristic sites are not accurate exists in either 3D or 2D system.Therefore,it is very necessary to look into the new measurement methods.Problems that the interference of irrelevant background to the extraction process and the inaccurate measurement results of some key characteristic sites exists in the original 2D non-contact body measurement system of our laboratory.In view of these problems,we made a series of improvement research in this paper.We eliminated the interference of irrelevant background to the extraction process with the help of the automatically crop of the 2D human body images.Based on the human body size collected from the 3D non-contact measurement system,we established the fitting rules of relevant parts to improve the accuracy of the measurement results of the 2D non-contact measurement system.Optimized the 2D non-contact measurement system to make the measurement process more convenient and improve the measurement accuracy of the 2D system.The main work of this paper includes: (1)AnthropometryManual measurement,2D and 3D non-contact measurement were used to measure the body sizes of 300 groups of young female.The analysis of the 3D human point-cloud data was carried out to measure the shoulder width and neck girth associated eigenvalues with the reverse engineering software named IMAGEWARE,including the seven characteristic values as straight shoulder width,curve shoulder width,back enter angle,neck width,neck thickness,and neck girth.(2)Classification fitting of shoulder width and neck girthBased on the data processing,the shoulder width and neck girth associated eigenvalues were classified and fitted to get the fitting formulas of them.On the basis of the data measured in the 3D human point-cloud,the fitting formulas were verified and the verification results showed that the stability and accuracy of the algorithms fitting results are all improved.(3)Batch automatic cutting of 2D imagesSelected some of the representative 2D images to determine the location and size of the clipping window with the manual method and then clipping all the images according to the clipping window.Browsed all the cropped pictures to find out whether the clipping position is correct and the size is appropriate.Verify by repeated operation to ensure the information integrity of the cutting results.If there were excessive cutting pictures we should redefine the clipping window to ensure the information integrity of the cutting results.Finally,the cutting experiment was carried out to prove that the cutting algorithm is viable.(4)System optimization and effectiveness verificationThe clipping algorithm and the fitting formulas were combined with the original 2D non-contact measurement system of our laboratory to obtain a new system.Then we used the new system after optimization to measure the shoulder width and neck girth and verified the effectiveness of the system optimization through the analysis of the final results.In the verification process of shoulder width measured in the new system,compared with the results of shoulder width in the manual measurement,the proportion of shoulder width which had the error absolute value within 2cm increased by 11 percent and the proportion of shoulder width which had the error absolute value within 1cm increased by 8 percent.The mean absolute value of the differences between the shoulder width measured in the new system and in the manual measurement reduced by 0.21 cm and the standard deviation reduced by 0.04 cm.In the verification process of neck girth measured in the new system,compared with that in the manual measurement,the proportion of neck girth which had the error absolute value within 1cm increased by 6 percent.The mean absolute value of the differences between the neck girth measured in the new system and in the manual measurement reduced by 0.14 cm and the standard deviation reduced by 0.047 cm.Results show that the stability and accuracy of the predictive value of shoulder width and neck girth in the new system have been increased. |
