Point cloud upsampling algorithms can improve the resolution of point clouds and generate dense and uniform point clouds, and are an important image processing technology. Significant progress has been made in point cloud upsampling research in recent years. This paper provides a comprehensive survey of point cloud upsampling algorithms. We classify existing point cloud upsampling algorithms into optimization-based methods and deep learning-based methods, and analyze the advantages and limitations of different algorithms from a modular perspective. In addition, we cover some other important issues such as public datasets and performance evaluation metrics. Finally, we conclude this survey by highlighting several future research directions and open issues that should be further addressed.
This paper presents a method to simultaneously measure the thickness and refractive index of the thermally grown oxide (TGO) in thermal barrier coating (TBC) by using a reflective terahertz time-domain spectroscopy (THz-TDS) system. First, an optical transmission model of THz radiation in the multilayer structure of TBC is established. Owing to the different structures of TBC before and after forming the TGO layer, two different transmission models are established, respectively. Then, the experimental signals from the samples after different thermal cycles are obtained by the THz-TDS system. By fitting the experimentally measured reflected THz signals from the TBC samples to the proposed optical model using an optimization algorithm, the thickness and refractive index of the TGO are determined. In this work, four samples with different thicknesses of TGO layers are analyzed. The results of thickness of TGO layer are verified by SEM observation, and a reasonable agreement is obtained.
X-ray diffraction can non-destructively reveal microstructure information, including stress distribution on thermal coatings of aeroengine blades. In order to accurately pinpoint the detection position and precisely set the measurement geometry, a 3D camera is adopted to obtain the point cloud data on the blade surface and perform on-site modeling. Due to hardware limitations, the resolution of raw point clouds is insufficient. The point cloud needs to be upsampled. However, the current upsampling algorithm is greatly affected by noise and it is easy to generate too many outliers, which affects the quality of the generated point cloud. In this paper, a generative adversarial point cloud upsampling model is designed, which achieves better noise immunity by introducing dense graph convolution blocks in the discriminator. Additionally, filters are used to further process the noisy data before using the deep learning model. An evaluation of the network and a demonstration of the experiment show the effectivity of the new algorithm.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.