摘要
本文针对实验力学云纹干涉图像去包裹处理中容易导致拉线的问题对RIPI(区域识别、划分和积分)去包裹法进行了研究。文章首先建立了RIPI的基本原理并对其进行了数学推导。在此基础上,本文给出了RIPI分区、求解、线性变换的三步实用算法,其中引入了数据结构中树的先根遍历原理用于分区,提出了区域积分的思想进行相位求解。本文最后将RIPI应用于云纹干涉实验图像处理并与其它方法进行了比较,应用和比较结果表明:着眼于整体又尊重局部的RIPI对云纹干涉实验图像特别是质量差的图像处理效果比较理想。
This study investigates a new 2-dimensional phase unwrapping method named RIPI (Regional Identification, partition and Integral phase unwrapping method), which is aiming at the problems of 'line-draw' and loss of useful data while phase unwrapping in processing moiré interferometry experiment images. Firstly, the principle of RIPI is established and deducted mathematically. Then, the 3-steps practicable algorithm of RIPI, Partitioning, Solving and Linear Transformation, is given. In the first step of this algorithm, the technique of prefix traversal of expression tree in computer data structure is introduced and applied to partition the image into regions. In the second step, the thought of regional integral is presented and formulized to solve the parameters of every pixel that is named phase-offset. At the end step, all the phases of the image are linearly transformed into [-π,π]. Eventually, RIPI is demonstrated by processing actual moiré interferometry experiment images. In contrast to earlier algorithms, RIPI can pay equal attention to entire and local information of the phase images, and it can be free from the problems of 'line-draw' and loss of useful data. It is proved that RIPI phase unwrapping method is enough efficient for the phase unwrapping of the moiré interferometry images, especially of the ones with poor quality.
出处
《实验力学》
CSCD
北大核心
2004年第2期129-135,共7页
Journal of Experimental Mechanics
基金
国家自然科学重点基金(10232030)