A Component Decomposition Model for 3D Laser Scanning Pavement Data Based on High-Pass Filtering and Sparse Analysis

Area of Science:

• Civil Engineering
• Geospatial Data Analysis
• Pavement Engineering

Background:

• High-precision 3D laser scanning provides rich pavement data.
• Automatic extraction of pavement indicators is crucial for maintenance and management.
• Existing methods struggle to extract multiple pavement indicators simultaneously.

Purpose of the Study:

• To propose and validate a 3D Pavement Components Decomposition Model (3D-PCDM).
• To enable simultaneous extraction of various pavement indicators from 3D data.
• To enhance pavement maintenance and management through accurate data analysis.

Main Methods:

• Analyzing frequency and sparse characteristics of pavement distresses and performance indicators.
• Decomposing 3D pavement profiles into sparse (x), low-frequency (f), and vibration (t) components.
• Employing a high-pass filter for f separation and total variation de-noising for x and t separation.

Main Results:

• The 3D-PCDM successfully decomposes 3D pavement profiles into distinct components.
• Decomposed sparse component (x) accurately characterizes location and depth of cracks and road markings.
• Extracted sparse components achieved an accuracy exceeding 92.75% in experimental validation.

Conclusions:

• The 3D-PCDM is effective for comprehensive pavement information extraction.
• The model enables accurate identification of pavement distresses and deformations.
• This approach significantly advances automated pavement assessment and management.