Automatic body feature extraction from a marker-less scanned human body

Abstract

In this paper, we propose a novel method of body feature extraction from a marker-less scanned body. The descriptions of human body features mostly defined in ASTM (1999) and ISO (1989) are interpreted into logical mathematical definitions. Using these significant definitions, we employ image processing and computational geometry techniques to identify, automatically, body features from the torso cloud points. We have currently extracted 21 feature points and 35 feature lines on the human torso; this number may be extended if necessary. Moreover, less than 2 min processing time is taken for body feature extraction starting from the raw point cloud. This algorithm is successfully tested on several Asian female adults who are aged from 18 to 60.

Introduction

Anthropometry is an important issue in the field of human factor related production. A precise sizing system provides a useful foundation for the manufacture of daily commodities. Among these, the design of human apparel, such as clothes, glasses, hats and footwear is most important, needing precise civilian anthropometric data. Body dimensions are usually tape-measured by tailors. The accuracy of measurement is affected by the expertise of the operators and the cooperation of the person to be measured. Conducting nationwide large-scale anthropometry is time consuming and tedious. There is great need for an automatic anthropometry system. Recently, the 3D body scanner has become a notable tool for anthropometry. Anthroscan [1], BodyShape [2], Cyberware [3], Gemini [4], Hamamatsu [5], Inspeck [6], TC² [7], TriForm [8], Vitus [9] are some examples. However, the raw data taken from such 3D scanners cannot be readily used by the industry because the scanning points contain little meaningful information. The key points in solving the problem are data extraction, feature identification, and data symbolization. Therefore, much research effort has been put into the post-processing of the raw data. Nurre separated the data into six regions by finding cusps on every slice of discrete points [10], [11]. Ju et al. segmented the body into 5 parts and determined the feature points layer by layer by computing its circumferences [12]. Wang el at. used fuzzy logic to recognize features in unorganized cloud points [13]. Pargas used the 3D scanner for measurement of body dimensions in the garment industry [14]. In the famous CAESAR project executed in Europe, Robinette et al. used a 3D body scanner to collect the dimensions of the body [15]. Prior to scanning, reflective markers were attached to the anatomical landmarks of the subject. Then the positions of the landmarks were determined semi-automatically. They used a neural net to identify the positions of the landmarks, by means of feature points [16]. Ashdown constructed a sizing system using the data from a 3D body scanner [17]. Wang developed an algorithm to extract key features on the human body, then built a set of parametric surfaces to represent the scanned subject [18]. Simmons compared different 3D scanners and concluded that a standard feature terminology and common feature recognition software is needed for various scanners [19]. Turning the scanned data into useful information for design purposes is still a long way off.

In this paper, we propose a novel method for body feature extraction from the scanned human body. The semantic definitions of body features found in ISO 8559 were interpreted into a series of mathematical definitions [20]. A total of 21 feature points and 35 feature lines on the human torso were identified. Each feature stands for an important landmark for garment making or the ergonomics industry. Contrary to previous studies, markers were not needed in this study. The features were identified by geometric properties and common proportions, thus eliminating the variations due to different operators. The overall goal of this research is to generate, automatically, a digital mannequin from the scanned body. With the anatomical features embedded in the digital mannequin, the automatic anthropometry system can be simply applied in the garment design industry, for example, as well as in ergonomic applications. Fig. 1 shows the workflow for digital mannequin generation. First, a subject is scanned by a full body scanner. The generated point cloud is aligned along its principle axes and segmented into 5 major parts: arms, legs, and a torso-and-head segment [21]. Then, the torso is encoded into a ranged map, in order to eliminate noise and fill up the void inside. Geometric features are easily discovered in the encoded image. Finally a parametric triangular tessellation is generated which retains every feature in it. This paper focuses on the investigation of feature extraction as a base for the next stage in tessellation and automatic anthropometry.

This paper is organized as follows. In Section 2, a detailed description is given for the techniques used for feature identification. In Section 3, we describe the methods for data alignment, segmentation and coded image representation. Next, we describe the definitions and search procedure for each feature on the torso. Finally, we discuss the results of our approach.