Illumination Pattern Optimization for Defect Inspection Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing inspection methods struggle to distinguish between defects and non-defects on workpieces due to variations in appearance and surface states, leading to missed defects or false positives, as they primarily focus on emphasizing defects without considering variations in non-defective products or individual workpieces.
Innovation Solution
An inspection device and method that sets image areas for non-defective and defective products, using a first index to maximize differences and a second index to minimize variations within the same label, optimizing illumination patterns to enhance defect recognition while reducing background variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If an illumination pattern is optimized to emphasize defects in workpieces, then defect detection capability is improved, but variations in non-defective products and individual workpieces are also emphasized, leading to false positives and missed defects
Solution Approach 1:
The patent segments the evaluation process into two distinct indices: a first index for evaluating defect detection capability and a second index for evaluating variations in non-defective products. By separately calculating and balancing these two indices, the system can optimize illumination patterns that emphasize defects while simultaneously suppressing background variations, thus resolving the contradiction between detection capability and inspection accuracy
Solution Approach 2:
The patent changes the evaluation parameters from a single defect-emphasis metric to a dual-parameter system that independently evaluates both defect detection performance and background variation suppression. This parameter transformation allows the optimization process to balance both objectives, preventing false positives while maintaining high defect detection capability
2Measurement precision
If manual optimization of illumination patterns is performed to reliably detect defects, then detection performance is improved, but the number of processes and time required significantly increases
Solution Approach 1:
The patent enables the system to automatically evaluate and optimize illumination patterns using the dual-index method without requiring manual intervention. The computer automatically calculates the first index for defect detection and the second index for background variations, then iteratively optimizes the illumination pattern based on these evaluations, significantly reducing the time and effort compared to manual optimization while maintaining high detection performance
Solution Approach 2:
The patent implements a feedback mechanism where the evaluation results of the first and second indices are used to iteratively adjust and optimize the illumination pattern. The system continuously evaluates the current illumination pattern's performance on both defect detection and background suppression, then uses this feedback to refine the pattern until optimal performance is achieved, automating what previously required repeated manual adjustment and review
Data Source
Figure 1A
Figure 1B
Figure 2
AI summary
This inspection device comprises: an image-capturing unit which obtains an evaluation workpiece image that is captured with a plurality of predetermined illumination light emission patterns in order to obtain an optimized illumination light emission pattern with which the influence of non-defective product variations of a workpiece can be suppressed while defects in a workpiece are emphasized; an image area setting unit which sets a plurality of image areas associated with a plurality of different labels; an optimization calculation unit which generates a first index of which the output value increases as a difference between an image area associated with a non-defective label and an image area associated with a defective label increases, and a second index of which the output value increases as a difference or contrast between the image areas associated with the non-defective label increases, and calculates an illumination light emission pattern for inspection such that the first index becomes larger and the second index becomes smaller; and a determination unit which performs image processing on an inspection workpiece image captured with the illumination light emission pattern for inspection, and determines a pass/fail label of a workpiece to be inspected.