Image Processing Apparatus for Protrusion Defect Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for inspecting industrial products with protruded/recessed shapes along container edges struggle to accurately detect protrusion/recess defects while allowing for individual variations in the slope of inclined surfaces.
Innovation Solution
An image processing apparatus and method that obtain a map representing the distribution of information related to protrusions and recesses from captured images, and determine the presence or absence of protrusion/recess defects at the boundary between inspection and non-inspection regions based on the uniformity of this information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a protrusion/recess defect is detected based on an angular difference between normal vectors of inspection object and reference object, then defect detection capability is improved, but individual variations in slope cause false defective product determination
Solution Approach 1:
The inspection region is divided into multiple sub-regions (first through fourth sub-regions) along the boundary portion. By segmenting the boundary inspection into distinct zones and analyzing normal vector distributions separately in each zone, the system can detect local slope variations while distinguishing them from individual product characteristics, thereby maintaining defect detection precision without false positives from natural variations.
Solution Approach 2:
The inspection method applies different analysis approaches to different parts of the boundary portion. Specifically, the first and second sub-regions are analyzed separately from the third and fourth sub-regions, allowing the system to account for local slope characteristics that vary by position while still detecting actual defects. This local differentiation resolves the contradiction by adapting the inspection criteria to local conditions.
2Area of stationary object
If the inspection region extends to the boundary portion, then defect detection coverage is improved, but false detection of normal variations as defects increases
Solution Approach 1:
The boundary portion is segmented into first through fourth sub-regions with specific analysis methods applied to each. This segmentation allows the inspection to cover the entire boundary area while applying position-specific evaluation criteria that account for expected variations in different zones, thereby maintaining both comprehensive coverage and detection accuracy.
Solution Approach 2:
The inspection method changes the evaluation parameters based on position within the boundary portion. By analyzing normal vector distributions differently across various sub-regions (e.g., comparing adjacent sub-regions vs. comparing with reference object), the system adapts its measurement criteria to local characteristics, enabling accurate defect detection across the full inspection area without false positives from normal variations.
Data Source
AI summary
An image processing apparatus includes an obtaining unit configured to obtain a map representing a distribution of information related to protrusions and recesses based on a plurality of captured images obtained by imaging an inspection object, and a determination unit configured to determine presence/absence of a protrusion/recess defect in a boundary portion between an inspection region and a non-inspection region in the map based on a uniformity of the information in the boundary portion.