A Survey on Surface Reconstruction Techniques for Structured and Unstructured Data

Surface reconstruction of real-world objects is a commonly discussed topic in reverse engineering. Generally, 3-D scanning technologies are used to scan the objects through multiple angles and represent them using point cloud. The point cloud can be either in structured or unstructured form which may contain problems such as noise, outliers and incomplete points. The point cloud is considered as unstructured form when it does not contain any connectivity information between adjacent points and structure information. Various types of surface reconstruction techniques are proposed to overcome the problems of point cloud and the limitations of existing techniques. Besides, soft computing techniques are also employed to enhance the performance and overcome the downsides of existing techniques. Therefore, the objective of this paper is to conduct a survey towards the existing techniques in the surface reconstruction on structured or unstructured data. Generally, this paper will only focus on the interpolation and approximation techniques, learning-based techniques, and soft computing techniques. Based on the analysis, it shows that learning-based techniques performed better compared to other techniques as they are able to handle the problem of unstructured point clouds. It can also form as hybrid techniques by integrating with other techniques which can improve its accuracy. The outcome of this paper can be used to assist the researchers in understanding and finding suitable surface reconstruction techniques in representing the objects and solving their case studies.