一种新的质量图引导的相位展开算法

A New Phase Unwrapping Algorithm Based on Quality Map Guidance

相位展开主要用于车轮探伤时提取车轮外形信息,不同算法展开后相位的质量也是不同的。介绍了传统的质量图引导的相位展开算法,针对该算法在处理问题时的不足,对其进行改进,得到了一种新的质量图引导的相位展开算法。该算法将像素点二阶差分值的倒数定义为质量点,通过将相邻像素点进行平均,得到全新的质量值,引导相位展开。通过比较枝切法、质量图导向法、最小二乘法以及本文的改进算法四种相位展开算法的仿真实验,结果表明该算法展开误差小,计算速度快,能够有效处理相位展开中不连续的问题。最后通过车轮实验验证了结果。

Phase unwrapping algorithm is mainly used to extract the profile information of wheels in flaw detection. The quality values of phases are also different for different algorithms. The traditional quality-map-guided phase unwrapping algorithm is introduced firstly in this paper. A new quality-map-guided phase unwrapping algorithm is proposed aiming at the shortages of the traditional. This algorithm defines the reciprocal of the second order difference value of pixel point as the quality point. By averaging the quality values of adjacent pixel points, a new quality value is obtained to guide the phase unwrapping. The simulation results of four phase unwrapping algorithms- Branch-cut method, quality map guidance, least squares method, improved algorithm presented in this paper are compared and show that the improved algorithm has smaller error and faster calculating speed, and can deal with the problem of discontinuity in phase unwrapping effectively. Finally, the practical wheel experiments are completed and also verify the conclusion.