A cumulative projection-based automatic registration method for mobile laser scanning data. Issue 1 (2nd January 2019)

• Record Type: Journal Article
• Authors: Sui, Yinling; Qin, Zhiyuan; Tong, Xiaochong; Li, He; Lai, Guangling
• Is Part Of: Remote sensing letters. Volume 10:Issue 1(2019)
• Page Start: 86
• Page End: 94
• Publication Date: 2019-01-02
• Subjects: Remote sensing -- Periodicals; Remote sensing; Periodicals; 621.3678
• DOI: 10.1080/2150704X.2018.1523582
• Languages: English
• ISSNs: 2150-704X

Abstract

ABSTRACT: This paper presents an automatic method for global alignment of mobile laser scanning (MLS) data in urban environments. Most of the existing algorithms either highly time consuming or invalid because of the data sets are always huge and the overlapping part always less than half. To address these problems, we propose an approach which consists of two consecutive stages. The first stage is cumulative projection transformation. In this stage, we convert the 3D-point clouds into 2D images using cumulative projection transformation. As a result, we could use the image-matching methods to fulfil the question. The second stage is image matching based on line segment. In the projection images, there are plenty of shape features, so we use the line segment detector to extract line segments and construct line-junction-line (LJL) structures, and then match the line segments and structures based on the line segment matching method. Experimentation on several data sets with tens of millions points from large regions of different cities demonstrates that our approach decreases the complexity and time-consumption significantly, and applies more accurately.