Research On Non-contact Measurement Of Young Men's Bodies Based On Digital Images

Non-contact measurement technology has been applied in human body measurement. Non-contact measurement instruments were introduced by higher academy or academic institutions from abroad. However, those instruments were not widely applied in China because of the differences in bodily form between Chinese and foreigner as well as higher price. Hence, it is important to develop non-contact measurement which can be operated easily and low costly.In this thesis, non-contact measurement system was developed based on extraction of silhouettes from digital images, determination of feature points as well as prediction of height, length or circumference. The advantage of this system was that it can be applied conveniently in the apparel industry to meet the requirement of measure to made and quick response. The main research work is presented as follow.Young men were selected as subjects. The measures on bodily form were obtained by using photogrammetry, traditional manual ways and non-contact automatic methods.Angles of bodily form, such as shoulder slope angle, dorsal angle, hip bulge angle, back enter angle and body lateral angle were used to describe the characters of bodily form. The correlation analysis among those angles was carried out to obtain feature angles. The cluster analysis was done based on minimal distances. The number of cluster was determined based on multivariate analysis of variance. In addition, the discriminants were developed to estimate new sample.Human body images taken by digital camera were focused. The images were further processed by using the algorithms of color segmentation, best threshold segmentation, filling holes, opening and pixel connectivity, and the silhouettes were extracted from the processed images. The extracted smooth and closed silhouettes confirmed that the proposed method was effective to process human body images.The extracted silhouettes of human body images were focused. The feature points on the silhouettes were determined based on the characters of bodily form. Then, body height, arm length, shoulder width, neck width, neck thickness, neck base width, neck base thickness, chest width, chest thickness, waist width, waist thickness, belt width, belt thickness, hip width and hip thickness were obtained.Cervical height, neck circumference, neck base circumference, chest circumference, waist circumference, belt circumference as well as hip circumference were predicted by using trained Back-Propagation Neural Network.The interface of non-contact measurement system was designed. The feasibility of this system was tested. The obtained measures confirmed that the developed system was effective to measure young men's bodies.