Micro-Defect Detection on TWS Earphones using Photometric Stereo
-
-45%False Rejection Rate
-
< 500msInspection Cycle Time per unit
-
0.02mmMin. Defect Depth
-
ObjecEarphone
-
RegionTaiwan
-
ModelDS081
Content
-
Project Background
-
Challenges
-
The Solution
-
Inspection Results
-
Client Value
-
Project Background
In the manufacturing of high-end consumer electronics, such as TWS (True Wireless Stereo) earphones, cosmetic quality is a critical factor influencing brand perception. However, automated inspection is notoriously difficult due to the product's curved geometry, highly reflective surfaces, and monochromatic (often pure white) designs.
-
Challenges
Our client faced significant challenges with traditional 2D machine vision solutions:
Extremely Low Contrast: Under standard single-source lighting, shallow scratches or micro craters on white surfaces blend into the background, making them nearly impossible to detect.
Feature Confusion: Conventional imaging struggles to distinguish between harmless surface stains (2D color changes) and actual physical damage (3D depth changes), leading to high false rejection rates.
-
The Solution
To overcome these limitations, we deployed a high-precision inspection solution based on Photometric Stereo technology. Unlike standard imaging, this system utilizes advanced computational photography to separate shape from texture.
Technological Highlights:
Multi-Spectrum & Multi-Segment Illumination:The system utilizes a specialized lighting controller to rapidly capture multiple images under varying illumination angles and wavelengths.
Surface Normal Analysis:By processing the sequence of images, the algorithm reconstructs the Surface Normal Map (X/Y direction). This effectively generates a slope map of the surface, completely isolating physical defects from surface color or pattern interference.
-
Inspection Results
The implementation of this solution achieved distinct visibility of micron-level defects on complex curved surfaces.
Standard Imaging (Left): Under normal lighting conditions, the defects are virtually invisible to the camera, resulting in potential quality escapes.
Photometric Stereo / Normal Map (Right): After algorithmic processing, the background noise is filtered out, revealing the defects clearly:
Scratch: Revealed as a distinct linear depth feature.
Crater: Identified clearly as a localized surface depression.



-
Client Value
The successful deployment of this system delivered immediate benefits to the production line:
Enhanced Detection Rate: Successfully identified low-contrast defects (shallow scratches/dents) that were previously undetectable.
Reduced False Positives: accurately distinguished between cosmetic stains and physical damage, saving costs on unnecessary scrap.
Versatility: The solution is highly adaptable and has since been applied to other components, including charging cases, smartphone back covers, and precision injection-molded parts.
CONTACT US
-
Trusted by Global Industry Leaders
Proven in high-end production lines of top-tier companies.
MTBF ≥100,000 h certified. Get your rush orders covered in 72h.
-
Unmatched Service & Reliability Guarantee
Speedy response within 4h, solutions within 7 days, everything from fault diagnosis, part replacement to validation report.