一种新的质量图引导的路径预测区域增长算法  

A Region-growing Phase Unwrapping Algorithm Based on a New Quality-guided Map

作　　者：刘颖[1] 张瑞峰[1] 陈相舟 李锵[1] 纪鑫[2] 汪洋[2] 

机构地区：[1]天津大学电子信息工程学院,天津300072 [2]中国电力科学研究院信息通信研究所,北京100192

出　　处：《光电工程》2016年第1期36-41,共6页Opto-Electronic Engineering

基　　金：国家自然科学基金资助项目(61471263)

摘　　要：针对五步相移干涉测量,提出了一种新的质量图用于引导路径预测区域增长算法。首先基于分支设置处理残差点,直至相位图中所有残差点都被平衡。然后将包裹相位图按质量值分割为若干区域,把分支对应的质量值设为最低,按质量值由高到低顺序对每一区域的相位进行路径预测区域增长方式的去包裹,若区域间有交界,则通过调整偏移量的方法进行融合处理。软件仿真和实验验证结果表明,与传统区域增长算法相比,新算法准确度更高,计算速度更快。For the measurement of five step phase shift optical interferometry, a new quality-guided map is proposed to be used in the region growing algorithm. Firstly, all the residues in the wrapped images are balanced based on the placement of the branch cuts. Then, the wrapped phase image is divided into some regions based on different quality values of the pixels, and the minimum quality value is assigned to pixels that belong to the branch cuts. The new quality-guided map is used to guide the region-growing unwrapping algorithm based on predictions from the highest quality values to the lowest. If the border area is overlapping, adjust the offset to merge. The proposed algorithm works fast and is robust against noise, as demonstrated in experimental and simulated data, compared with the traditional region-growing unwrapping algorithm based on predictions.

关 键 词：区域增长 相位去包裹 质量图 五步相移 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]