Research On Quality Maps Used In Two Dimensional Phase Unwrapping

Two-dimensional phase unwrapping is the key technique for phase measurements, such as optical interferometry and magnetic resonance imaging etc. In these applications, the physical quantities to be measured are represented by phase, which is normally in the interval (-π,π] due to the mathematical arctangent function. In order to recovering the true continuous phase values, a phase unwrapping process is needed to recover the true phase values. Phase unwrapping is the task of recovering the true phase values from the wrapped phase image. In many phase unwrapping algorithms, a quality map is used for completing the phase unwrapping process. Quality maps are arrays of values that define the quality of each pixel of the given wrapped phase data. Whether the unwrapped result is correct, to a large extent, depends on whether the quality map can truly reflect the wrapped phase quality. The existing quality maps have some shortcomings and may not reflect the phase data quality accurately due to noise and undersamping etc, which may lead to unreliable unwrapped results. To solve the problem, the quality maps for two dimensional phase unwrapping are deeply studied and discussed in this thesis.After detailed description of the principles and the shortcomings of existing quality maps, innovative ideas of weight gradients and correlation are introduced to construct new quality maps. The weighted phase gradients based quality map can be used to optimize quality maps based on phase gradients. The weighted phase gradients based quality maps which aim to overcome the shortcomings of traditional phase gradients based quality map, utilizes the gradient information of neighborhood pixels, and thus can reflect the wrapped phase data quality more accurately. Proper weighted gradients can be chosen to avoid or reduce the impact of noise and undersamping etc, and optimize the results of phase unwrapping. A phase gradient correlation based quality map is a new quality map constructed by the phase gradient correlation coefficient of each pixel within its neighborhood. The quality map can effectively distinguish between noise area and the reliable area with large phase gradients, and can reflect the reliability of the wrapped phase data. Based on the above two quality maps, the weighted phase gradients correlation based map which combines the advantages of the both quality maps is proposed. The quality map can improve the unwrapped results significantly. To evaluate the performance of quality maps, a new unwrapped results evaluation method is also proposed. The unwrapped results show that the three new quality maps can help to enhance the reliability of phase unwrapping and can be considered as reliable phase-quality indicators.