A novel capacity expansion and recognition acceleration method for dot-dispersing coded targets in photogrammetry
A novel capacity expansion and recognition acceleration method for dot-dispersing coded targets in photogrammetry
复制标题
DOI：
10.1088/1361-6501/ac89a3
复制
发表时间：
2022-08
影响因子：
2.4
通讯作者：
M. Wang;Yu Guo;Qiang Wang;Yang Liu;Jiaxin Liu;Xinmin Song;Guo Wang;Hu Zhang
中科院分区：
文献类型：
--
作者：
M. Wang;Yu Guo;Qiang Wang;Yang Liu;Jiaxin Liu;Xinmin Song;Guo Wang;Hu Zhang
在近景摄影测量中，已经开发了一些改进的设计来提高类施耐德同心圆编码目标的编码容量。然而，另一种典型的类型，服部类散点编码目标(HCT)，很少被扩展。在本文中，我们给出了一种用于视频同时三角测量和后方交会系统中的非公共HCT的详细解决方案。在此基础上，提出了一种多模板、简化的P2不变量的改进HCT算法(AHCT)，实现了容量扩展和识别加速。首先利用交比不变量进行模板识别，然后利用仿射变换进行解码。在模板识别过程中，设计了多个模板以扩展编码容量，并采用简化的P2-不变量来加速识别。在0°~80°的不同拍摄角度、较小的编码目标、不同大小的混合编码目标和较大的室外场景下进行了实验，结果表明，AHCTS不仅保持了编码目标的稳定性，而且在较快的识别率下实现了编码目标的容量增长14.6%。大容量的AHCT可以应用于规模较大或结构复杂的场景，在实时动态域中可以考虑加速优势。
In close-range photogrammetry, some improved designs have been developed to increase the coding capacity of Schneider-like concentric circular coded targets. However, another typical type, Hattori-like dot-dispersing coded targets (HCTs), is rarely expanded. In this paper, we give our detailed solution to recognize a kind of non-public HCT used in video-simultaneous triangulation and resection systems. Then, an advanced HCT (AHCT) with multiple templates and simplified P2 -Invariant is proposed, which achieves capacity expansion and recognition acceleration. First, the cross-ratio invariant is used in template recognition, and then the affine transformation is used in decoding. During the template recognition process, multiple templates are designed to expand the coding capacity, and a simplified P2 -Invariant is adopted to accelerate the recognition. Experiments are carried out with different shooting angles ranging from 0° to 80°, with smaller coded targets, with mixed coded targets of different sizes and with large outdoor scenes, and the results show that AHCTs can not only preserve the stability of the HCTs, but also achieve the capacity growth of coded targets at a faster recognition rate by 14.6%. The AHCTs with large capacity can be applied in scenarios with a large size or complex structure, and the acceleration advantage can be considered in real-time dynamic domains.