Semantic air segmentation in core with noise reduction and post-processing

This paper presents results on semantic air segmentation for digital concrete cores. Brightness normalization and noise reduction were used as image preprocessing. To increase the efficiency of segmentation, a method is proposed for segmentation in three directions, followed by summing the resulting...

Full description

Saved in:
Bibliographic Details
Main Authors: Mikhail I. Volkov, Polina S. Kargina
Format: Article
Language:English
Published: Tomsk Polytechnic University 2024-06-01
Series:Известия Томского политехнического университета: Промышленная кибернетика
Subjects:
Online Access:https://indcyb.ru/journal/article/view/53/48
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This paper presents results on semantic air segmentation for digital concrete cores. Brightness normalization and noise reduction were used as image preprocessing. To increase the efficiency of segmentation, a method is proposed for segmentation in three directions, followed by summing the resulting masks and using a median filter. The resulting image masks have a three-dimensional structure and can be used to generate a training dataset for 3D convolutional neural network architectures. The method proposed can be used as well to enlarge images in the training set if data labeling takes a long time.
ISSN:2949-5407