Vehicle Linear Element Defect Detection with Local Thresholding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image analysis algorithms for evaluating linear elements in vehicle components are limited by thresholding methods that fail to account for local image differences and are sensitive to positional tolerances, leading to inaccurate defect detection in elements like interior lighting and seams.
Innovation Solution
A method involving a camera-based image analysis that divides the evaluation region into sections, applies local thresholding, and determines brightness, thickness, and position-tolerant parameters to detect defects in linear elements, using a predefined ROI and weighted color channel superposition to create a grayscale image for efficient defect identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a single-pixel line polygon is used for evaluation, then the evaluation is simple and fast, but the evaluation is insufficient and highly sensitive to positional tolerances
Solution Approach 1:
The patent applies local quality by using a multi-pixel band instead of a single-pixel line, where each position along the linear element is evaluated using multiple neighboring pixels. This local expansion of the evaluation window provides robustness against positional variations while maintaining position-specific evaluation, thus resolving the contradiction between simple/fast evaluation and accurate/robust evaluation.
2Measurement precision
If a multi-pixel band is used for evaluation, then the signal noise is reduced and position tolerance is improved, but the evaluation becomes more complex and computationally intensive
Solution Approach 1:
The patent segments the evaluation process into distinct steps: defining the multi-pixel band, extracting pixel values, computing the average, and comparing with threshold. This segmentation of the algorithm into modular operations simplifies the overall complexity while achieving robust evaluation, thus resolving the contradiction between measurement precision and device complexity.
3Productivity
If global thresholding is applied to the entire image, then the processing is simple and fast, but local image differences are truncated and defects may be missed
Solution Approach 1:
The patent applies local quality by implementing local thresholding within the multi-pixel band at each position along the linear element, rather than using a single global threshold for the entire image. This allows the evaluation to adapt to local variations in lighting and background while maintaining processing efficiency, thus resolving the contradiction between processing speed and defect detection accuracy.
Data Source
Figure 1
Figure 2a~2c
Figure 3
AI summary
The disclosure relates to a method (100) and a testing system for detecting defects in linear elements of vehicle components. The method comprises the following steps: providing (101) a camera image of a vehicle component having a linear element; wherein the camera image comprises image elements with a respective brightness; determining (102) an evaluation region in the camera image having a plurality of rows, wherein the evaluation region comprises at least a portion of the linear element; determining (103) brightness values of the image elements of a respective row of the evaluation region; determining (104) a number of image elements of the respective row of the evaluation region whose brightness values lie beyond a brightness threshold, wherein the determined number corresponds to a thickness of the linear element in the respective row of the evaluation region;and detecting (105) a defect in the linear element of the vehicle component if the thickness of the linear element exhibits a fluctuation outside a tolerance range over the majority of the rows of the evaluation area;