Specular Surface Defect Detection Using Knife-Edge Vision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine vision systems face challenges in inspecting specular surfaces due to reflections that wash out defects, making it difficult to detect small slope differences, and existing techniques like dark field lighting have limitations, especially in environments with relative motion between the object and camera.
Innovation Solution
A system and method using a knife-edge technique where the camera aperture or an external device forms a physical knife-edge structure to block reflected rays from flat surfaces, allowing deflected rays from sloped defects to reach the camera sensor, with angled illumination and polarized light to enhance defect visibility.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If dark field lighting is used to highlight surface imperfections, then defect visibility is improved, but the technique has limitations in environments with relative motion between the object and camera
Solution Approach 1:
The patent replaces the mechanical/optical dark field lighting setup with a computational approach using deep learning neural networks. The system captures images under conventional lighting conditions and uses AI algorithms to distinguish defects from reflections, eliminating the need for specialized lighting geometries that fail with relative motion.
Solution Approach 2:
The patent changes the approach from modifying lighting parameters (dark field geometry) to modifying image processing parameters. By adjusting the neural network's analysis of image characteristics, the system adapts to various motion conditions without requiring physical reconfiguration of the optical setup.
2Device complexity
If conventional lighting is used for imaging, then the imaging setup is simple, but reflections wash out defects making them undetectable
Solution Approach 1:
The patent substitutes complex optical filtering mechanisms with a deep learning-based image processing system. Instead of using multiple light sources, polarizers, or complex aperture arrangements to separate defect light from reflection light, the system uses a neural network to computationally distinguish between the two based on learned patterns.
Solution Approach 2:
The patent introduces an intermediary layer of AI processing between the captured image and the final defect detection result. The deep learning model acts as a mediator that interprets the complex interaction between light, surface, and camera, extracting defect information that would otherwise be hidden in the reflected light.
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
Effectively detects and images defects on specular surfaces by blocking reflected light from flat areas and transmitting light from sloped defects, providing high contrast and accurate imaging of defects even on moving or layered surfaces.
Implementation Method 1
the camera aperture or an external device is set to form a physical knife-edge structure within the optical path that effectively blocks reflected rays from an illuminated specular surface
Implementation Method 2
The light is reflected by the (specular) region and continues converging to a spot either near the entrance aperture of the camera, or on the aperture stop
Implementation Method 3
The illuminator can include a linear polarizer that transmits polarized light to the surface of the object. The polarized light is reflected from the surface and into a crossed polarizer at the camera sensor/camera optics
Data Source
AI summary
This invention provides a system and method for detecting and imaging specular surface defects on a specular surface that employs a knife-edge technique in which the camera aperture or an external device is set to form a physical knife-edge structure within the optical path that effectively blocks reflected rays from an illuminated specular surface of a predetermined degree of slope values and allows rays deflected at differing slopes to reach the vision system camera sensor. The light reflected from the flat part of the surface is mostly blocked by the knife-edge. Light reflecting from the sloped parts of the defects is mostly reflected into the entrance aperture. The illumination beam is angled with respect to the optical axis of the camera to provide the appropriate degree of incident angle with respect to the surface under inspection. The surface can be stationary or in relative motion with respect to the camera.