Year

2013

Degree Name

Doctor of Philosophy

Department

School of Electrical, Computer and Telecommunications Engineering

Abstract

This thesis aims to improve the efficiency and accuracy of phase unwrapping for digital projection based three-dimensional fringe patterns profilometry (FPP). The proposed initiative is facilitated by the analysis of phase unwrapping principles and by the proposal of new phase unwrapping algorithms.

This thesis begins by tracing the history and development of FPP and then briefly introduces the principles, relevant conventional phase unwrapping methods, and typical applications of FPP. Focusing on phase unwrapping, the author theoretically analyses phase unwrapping errors in FPP and finds that these errors are triggered by disturbances on the phase map, which unavoidably makes conventional phase unwrapping methods unsuitable.

To solve the phase unwrapping problem, the author presents several approaches. The first approach is proposed to improve the efficiency of the quality-guided phase unwrapping method in terms of reducing the computational burden. This approach combines the conventional quality-guided method with a simple phase unwrapping algorithm. A quality threshold is used to classify pixels on the wrapped phase map into two types: high-quality (HQ) pixels corresponding to smooth phase variance and low-quality (LQ) pixels corresponding to rough phase variance. These two types of pixels are then unwrapped using different approaches. The HQ pixels are unwrapped by a simple phase unwrapping algorithm and the LQ pixels are recovered by a conventional quality-guided method. It is shown by both theoretical analysis and experiment that the proposed approach is able to unwrap complex phase maps with similar accuracy to the conventional quality-guided method, but with a much lower computational burden, and therefore, much faster.

The second approach is proposed to improve both phase unwrapping reliability and efficiency when the object surface has large depth discontinuities and shadows. Although phase unwrapping errors caused by large depth discontinuities and shadows are unavoidable, the phase unwrapping accuracy can be improved if the propagation of errors is prevented. Based on this idea, this approach blocks large depth discontinuities and shadows from participating in phase unwrapping and, as a result, has better results than the conventional method. An image of the object is retrieved from the fringe patterns, and the shadow areas and the large depth discontinuities are identified. When carrying out phase unwrapping, the edges and shadow areas are isolated, and the remaining areas are then unwrapped by means of an efficient group merging technique proposed. The experimental results show that the proposed method is able to deal with the phase unwrapping of objects with large depth discontinuities and shadows.

The third approach, a colour projection based method, is proposed to further improving phase unwrapping accuracy when an object has spatially isolated areas and distributed surface colour. Using this method, each fringe of the sinusoidal patterns is identified by a unique binary sequence. These sequences are encoded by a channel encoding scheme used in communications and the codewords are carried by binary coding stripes. With the colour projection technique, the sinusoidal fringes and the binary stripes can be projected simultaneously using red, green and blue (RGB) colours. The projected fringe patterns are captured and decomposed by an RGB 3-CCD camera into red, green and blue components. The wrapped phase map is obtained from the sinusoidal fringe patterns and the phase unwrapping is implemented based on the binary stripes. Compared with existing approaches, this approach provides reliable measurement of objects with spatially isolated areas and distributed surface colour.

Finally, the author concludes the thesis and proposes avenues for further research.

Share

COinS
 

Unless otherwise indicated, the views expressed in this thesis are those of the author and do not necessarily represent the views of the University of Wollongong.