激光与光电子学进展, 2018, 55 (2): 020301, 网络出版: 2018-09-10
采用改进KLT算法的标志点匹配方法 下载: 1229次
Marker Matching Method with Improved KLT Algorithm
相干光学 散斑测量 标志点匹配 改进加速稳健性特征算法 改进Kanade-Lucas-Tomasi算法 最大双向误差 coherence optics speckle measurement marker matching improved speeded up robust feature algorithm improved Kanade-Lucas-Tomasi algorithm max bidirectional error
摘要
在散斑视觉测量中,通常引入标志点以提高散斑的测量效率。针对传统标志点匹配过程中存在的匹配时间长、匹配准确率低等问题,提出了一种采用改进KLT(Kanade-Lucas-Tomasi)算法的标志点匹配方法。该方法在利用改进加速稳健性特征(SURF)算法对标志点进行检测以建立初始匹配点的基础上,采用改进的KLT算法实现标志点的匹配,并利用最大双向误差作为约束条件删除匹配过程中存在的误匹配点,以提高匹配的可靠性。最后,对机翼颤振测量中涂敷在机翼模型散斑区的标志点进行了匹配实验验证。结果表明,与传统的尺度不变特征转换(SIFT)与SURF匹配算法相比,所提方法在匹配时间上分别减少了75.9%和42.8%,在匹配准确率上分别提高了30.6%和22.2%。
Abstract
In speckle vision measurement, markers are usually used to improve the measurement efficiency. To overcome long matching time and low matching accuracy in traditional marker matching process, we propose a new method for marker matching by using improved Kanade-Lucas-Tomasi (KLT) algorithm. Firstly, we measure the marker to establish initial matching point based on the improved speeded up robust feature (SURF) algorithm. Secondly, we use the improved KLT algorithm to achieve the marker matching. Thirdly, we use the constraint condition based on max bidirectional error to delete the mismatched points and improve the reliability of maker matching. Finally, we make a matched experiment to verify the marker coated on the speckle region of the wing model during the wing flutter measuring. The results show that, compared with traditional scale-invariant feature transform (SIFT) and SURF matching methods, the proposed method reduces the matching time by 75.9% and 42.8%, and improves the matching accuracy by 30.6% and 22.2%, respectively.
于之靖, 马凯, 王志军, 吴军. 采用改进KLT算法的标志点匹配方法[J]. 激光与光电子学进展, 2018, 55(2): 020301. Zhijing Yu, Kai Ma, Zhijun Wang, Jun Wu. Marker Matching Method with Improved KLT Algorithm[J]. Laser & Optoelectronics Progress, 2018, 55(2): 020301.