Technical gazette, Vol. 32 No. 5, 2025.
Original scientific paper
https://doi.org/10.17559/TV-20250131002314
The Impact of Different Artificial Neural Network Structures on the Results of Preprocessing 3D Digitization Data
Slađan Lovrić
orcid.org/0009-0002-1719-4043
; Department of Production Engineering, Faculty of Mechanical Engineering, University of Tuzla, 75000, Tuzla, Bosnia and Herzegovina
*
Alan Topčić
; Department of Production Engineering, Faculty of Mechanical Engineering, University of Tuzla, 75000, Tuzla, Bosnia and Herzegovina
Edin Cerjaković
; Department of Production Engineering, Faculty of Mechanical Engineering, University of Tuzla, 75000, Tuzla, Bosnia and Herzegovina
* Corresponding author.
Abstract
Three-dimensional digitization (3D) is one way of collecting data in the formation of 3D models, for the development of new products, the reconstruction of existing ones, or general engineering practice. In order for the 3D model obtained by applying 3D digitization to be usable, the "raw" cloud of data generated by the 3D digitization process must be pre-processed before forming the 3D model. There are severel ways to preprocess data obtained from the 3D digitization process, and one of them is the application of artificial neural networks (ANN) in the process of preprocessing of 3D digitization data. The problems that arise when applying artificial neural networks (ANN) in the preprocessing process of 3D digitization data are in the structure of the ANN, which directly affects the preprocessing results. The aim of this paper is to highlight the impact of the structure of artificial neural networks on the preprocessing of 3D digitization data.
Keywords
artificial neural network (ANN); preprocessing; raw point cloud; three-dimensional digitization (3D)
Hrčak ID:
335086
URI
Publication date:
30.8.2025.
Visits: 540 *