The Study And Implementation Of Image-based Non-contacting Human Body Auto-Measurement System

Non-contacting human body auto-measurement technology obtains human body size quickly and exactly,to establish human body form database.It is applied widely not only for the study of human body measurement,the standard of clothing size and custom-made, human body modeling for all kinds of person,but also for providing the research resource about bodily form trends of our country and data sustaining by which experts design more comfortable,more customizedly daily necessities.In our country,it's very late for the study of Non-contacting human body auto-measurement.We introduce into advanced equipments from American,Europe,and Japan.These equipments are too expensive and not fit for our countries' bodily form. Meanwhile,the study is not applied for clothing manufacture and still for experiments. Aiming at above disadvantages,we develop a non-contacting human body auto-measurement system based on 2D images.The main study in this paper is that:(1)Because of the disadvantage that we can't obtain the human body's circumferences size from 2D images,we sort and analysis for circumferences,mapping the examples' circumference curves approximately.Then we make relationship using functional model between the circumferences and its width,depth,human stature and avoirdupois.Thus,it is easy to estimate the sizes by the models.Otherwise,we present a approach about based example functional model of human trunk for actual human body modeling.(2)Aiming at the contour points contain less information except their coordinates which is low level for images.So we need recongnise size points by searching all contour and it's low recongnition ratio.We present a method to divide human body contour based orientation.The method adds the geometry relationship attribute and object relationship sttribrte to divide contour points.Then combining two divisions,we can obtain the division of whole body contour by the knowledge of set theory.So the information contained in the contour points is extended from the low-level coordinate description to high-level description of human contour.The method can extract the specific parts of the profile, making the search of the size point conveniently,and improves the recognition accuracy effectively.(3) Environmental construction of measurement system and selection of equipment.In order to predigest the body contour collection and image processing,we consider the rationally of measurement background,clothing and light.Under this environment,the person will be comfortabal and his contour is extracted as clearly and conveniently as possible.In addition,according to the principle of computer vision,we should calibrate the camera and confirm the camera parameters,measurement of distance,emendating parameters of image distortion and so on.