A 3D camera is an imaging device that enables the perception of depth in images to replicate three dimensions as experienced through human binocular vision. Some 3D cameras use two or more lenses to record multiple points of view, while others use a single lens that shifts its position.

In fact, the first 3D whole body scanning system, the Loughborough anthropometric shadow scanner (LASS) [2] , was developed by the University of Loughborough (UK) in 1989 with the aim of conducting anthropometric surveys. The military industry was also using 3D scanning equipments, but its application was primarily ergonomics. Only lately, in the mid-nineties, whole body scanning systems began to be used also for uniform fit.

This marked the start of 3D body scanning for clothing industry, which was then finalized with the commercialization in 1998 by TC of the first 3D whole body scanner developed especially for the fashion industry. The methods and techniques for the digitization of the human body were in the two last decades continuously ameliorated, more efficient and powerful scanning systems were developed and new software tools were introduced for a more efficient use of the resulting data.

The number of available solutions increased and with the time, the different scanning systems profiled themselves more clearly for specific application fields. Moreover, with the possibility of a massive cost reduction given by new technologies, human body digitization became interesting to various fields of application, such as animation, computer games, art, sculpture, ergonomics, medicine, cosmetics, biometry and fashion

Complete full body scanning systems are usually accompanied by software solutions for the determination of body sizes from the 3D scan data. Such solutions are available since many years. The new advancements in this case are in the automation of the processing. Nowadays, typical software solutions measure completely automatically anatomical landmarks (about 90 points) and body measures (about 120). Moreover, the extracted measurements are often compatible with anthropometric and ergonomics norms (e.g. ISO7250, EN13402).

Two examples are shown in Figure 12, from Human-Solutions [12] and Cyberware [1]. Automation in the data processing is also accompanied by the advancements in the scanning process. In fact, recently have been developed fully autonomous body scanners, such as for 2 example the NX-16 of [TC] [3] (Figure 13). The entire scanning procedure and data processing are fully automatic and autonomous, so that there is no need of personnel to operate the scanner: the person operates alone the scanner and trig the scanning process by pressing a button located on holding levers inside the scanning cabin. Recorded messages and video presentations inside the cabin give additional information and instructions. The obtained 3D data is then processed fully automatically to extract body measures (as in 3.1). Figure 13: Autonomous full body scanning solution NX-16 of [TC] [3]. From left to right: the scanning cabin; inside the scanner; automatic data processing. The data resulting from a full body scanner is usually in form of a 3D surface model representing the scanned person. The main disadvantages of this representation form are the large amount of data required (several MB) and the impossibility to change the pose/posture of the body model; thus there is a

limited possibility to use 3D scan data for online applications and in application where the 3D model is animated. For these reasons, the use of parameterized 3D Avatars as descriptions of the scanned person has increased. In fact, Avatars can assume various poses and can be easily animated. Moreover, the size of data required to define a parameterized Avatar is very small (in the order of kB). However, a good approximation of the overall shape of the human body is important for applications as best-fit 2 or virtual-try-on (section 3.4). A recently developed solution (from [TC] [3].) allows indeed the generation of a personalized and parameterized Avatar starting from the 3D data obtained from a full body scanner. Figure 14 shows the different steps of the fully automatic process: a parameterized Avatar is modified so that its shape represents as closely as possible