LIDAR-Camera Crack Assessment for Real-World Dimension Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automated crack assessment methods using UAV-based lidar and image data face challenges in accuracy due to complex backgrounds and the inability to accurately quantify cracks in units of measurement without prior knowledge of image scales.
Innovation Solution
Integrate lidar data with image data to identify regions of interest (ROIs) and determine actual pixel sizes, using a convolutional neural network (CNN) to enhance crack detection and quantification by mapping pixels to corresponding points in point cloud data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated crack assessment methods are implemented, then productivity is improved, but measurement precision deteriorates due to complex backgrounds and inability to accurately quantify cracks
Solution Approach 1:
The patent segments the image into multiple patches and processes each patch independently through the CNN classifier. This segmentation approach allows the system to handle complex backgrounds by focusing on local regions, improving measurement precision while maintaining automated processing efficiency.
Solution Approach 2:
The patent introduces point cloud data as an intermediary to establish the relationship between image pixels and real-world dimensions. By mapping pixels to corresponding points in the point cloud, the system can accurately quantify cracks in units of measurement, resolving the precision issue while maintaining automated assessment.
2Ease of operation
If image-based crack detection is used, then ease of operation is improved, but measurement precision worsens due to lack of real-world dimension information
Solution Approach 1:
The patent merges image data with point cloud data to create a unified assessment system. By combining the ease of image-based detection with the dimensional information from point cloud data, the system achieves both operational simplicity and measurement precision, allowing cracks to be detected and quantified in real-world units.
Solution Approach 2:
The point cloud data serves as an intermediary that bridges the gap between image pixels and real-world dimensions. This intermediary enables the system to maintain the simplicity of image-based operation while achieving accurate quantification of crack dimensions in physical units.
3Measurement precision
If deep learning classifiers are applied to all image data, then measurement precision is improved, but use of energy increases due to processing large amounts of data
Solution Approach 1:
The patent segments the entire image into multiple smaller patches before applying the CNN classifier to each patch. This segmentation reduces the computational burden on the deep learning model, lowering energy consumption while maintaining high measurement precision through focused processing of relevant regions.
Solution Approach 2:
The patent applies the computationally intensive deep learning classifier only to specific patches that are likely to contain cracks, rather than processing the entire image uniformly. This partial application of the classification process reduces overall energy consumption while maintaining high precision where it matters most.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Accurately detects and quantifies cracks in real-world objects by filtering out background noise and determining real-world dimensions, improving the precision of crack assessment.
Implementation Method 1
the point cloud data is light detection and ranging (LIDAR) data captured by a LIDAR sensor
Data Source
AI summary
Embodiments automatically assess, e.g., quantify dimensions of, cracks in real-world objects. Amongst other examples, such functionality can be used to identify structural problems in bridges and buildings. An example implementation maps pixels in an image of a real-world object to corresponding points in point cloud data of the real-world object. In turn, a patch in the image data that includes a crack is identified by processing, using a classifier, the pixels with the corresponding points mapped. Pixels in the patch that correspond to the crack are then identified based on one or more features of the image. Real-world dimensions of the crack are determined using the identified pixels in the patch corresponding to the crack.