| The technology of optical three-dimensional (3D) sensing has wide prospects of application in industrial inspecting, dressmaking and biomedicine etc., and is popularizing. The methods of 3D sensing based on structured illumination, including Phase Measurement Profilometry (PMP), Fourier Transform Profilometry (FTP) and Modulation Measurement Profilometry (MMP) etc. are paied great attention in recent years.3D shape of objects is measured by analysing the deformation of fringe pattern casted on objects in PMP. The wrapped phase of fringe pattern is calculated from more than three frames of images containing phase shifted fringe pattern. Then, the continuous phase is obtained through unwrapping algorithm. Finally, 3D shape is reconstructed from continuous phase through proper algorithm. In this dissertation, the key technologies in PMP such as casting and acquiring fringe pattern, phase unwrapping, phase-height mapping and system calibrating etc. are studied. The factors affecting measurement accuracy in PMP are analysed. The data merge of two sensors and the problem of hidden face in visualization of 3D measurement data based on a computer generated hologram are discussed. The principal contributions of this dissertation are:1. It is revealed through experiment that the light field of DLP projector is temporal variable and the change is periodical due to the binary pulsewidth modulation used in DLP. The following conclusion is drawn. In order to obtain high measurement accuracy, the integrating period of camera must be integer times of the refresh period of projector in PMP.2. The phase unwrapping algorithm based on reliability is studied thoroughly. A fast algorithm using virtual queue is proposed. The speed of phase unwrapping is generally increased over 20 times. Especially, in the case that reliability value is dispersive, the speed is increased over 100 times. On the other hand, the success rates of phase unwrapping is high as well.
|
PDF Full Text Request