A Storage-Efficient Feature for 3D Concrete Defect Segmentation to Replace Normal Vector

TL;DR

This paper introduces a new relative angle feature for 3D concrete defect segmentation that reduces storage needs by 27.6% while maintaining performance, enabling more efficient processing on limited hardware.

Contribution

The study proposes a novel relative angle feature that replaces normal vectors, reducing storage and input complexity without sacrificing segmentation accuracy.

Findings

Achieved 27.6% storage reduction using the relative angle feature.

Maintained similar performance to normal vector-based models.

Enabled larger batch processing on resource-constrained hardware.

Abstract

Point cloud reconstruction of damage offers an effective solution to image-based methods vulnerable to background noise, yet its application is constrained by the high volume of 3D data. This study proposes a new feature, relative angle, computed as the angle between the normal vector of a point and the average normal vector of its parent point cloud. This single-dimensional feature provides directionality information equivalent to normal vectors for concrete surface defect characteristics. Through entropy-based feature evaluation, this study demonstrates the ability of relative angle to filter out redundant information in undamaged sections while retaining effective information in damaged sections. By training and testing with PointNet++, models based on the relative angles achieved similar performance to that of models based on normal vectors while delivering 27.6% storage reduction and 83% input channel compression. This novel feature has the potential to enable larger-batch execution on resource-constrained hardware without the necessity of architectural modifications to models.