Research On Automatic Human Body Measurement Based On Images

Human body feature point and feature size is the key to the costume design. Accurate extraction of body dimension is the basis of costume design, and feature point extraction directly affect body dimension measurement. Non-contact measurement based on images has been payed widely attention with its fast, convenient, low cost advantages. It is particularly applicable to virtual fitting system, clothing customization system based on network platform. Therefore, it is valuable and prospective to research human body automatic measurement method based on images. This topic focus on automatic extraction of feature points and body measurement based on front and side photos of human body obtained by an ordinary camera, now the main work summarized as follows:(1) This paper analyzes the research status of feature point extraction method and human body measurement techniques. According to the demands of the development of clothing design system, it offers the method of improving data measurement for obtaining the size of the human body based on images. This paper analyzes the performances of various edge detections, and selects Canny edge detection operator to meet the requirements of this topic research.(2) A method of human body feature points automatic extraction based on images is proposed. The proposed approach is tested on ten human subjects and the entire101feature points with clearly geometric properties are extracted automatically, including27points corresponding to the definitions of the landmarks about garment measurements. (3) An automatic human feature size extraction method based on detected feature points is proposed. The human body measurement methods are analyzesed in this paper. According to clothing standard,28practical and effective body sizes are obtained.