High-Speed Reflective Packaging Inspection
Detect errors in the packaging of razors, electric toothbrush heads, and other small items
Graphical programming environment for deep learning-based industrial image analysis
The stiff plastic packaging of many small personal care products, such as razors or spare electric toothbrush heads, is often transparent and reflective, making it difficult to inspect the packaging while confirming the package contents. Contents can vary in number, type, and arrangement.
When lit, the packaging creates bright reflections, leaving other parts of the image significantly darker. Increasing the brightness to bring out detail in the dark areas the consumer causes oversaturation from the reflective parts. Creating an evenly lit image for accurate inspection can take a great deal of effort and expertise.
Without such sophisticated lighting, traditional machine vision has great difficulty with images that have highly reflective parts. Often such packages must be manually inspected, which means slower speeds. If production speed means that only a sample of production can be fully inspected, errors can be missed.
Variable contents, such as different numbers, types, and colors of toothbrush heads, make it prohibitively time-consuming to program traditional rule-based machine vision to accurately inspect each new variant. Errors detected further down the line lead to increased waste, while products with undetected errors can lead to consumer dissatisfaction. .
Cognex vision systems with AI-powered tools and High Dynamic Range Plus (HDR+) technology are specifically designed to handle reflective inspection.
HDR+ technology uses an advanced algorithm that optimizes contrast in localized areas with just a single image, rather than the multiple images required by standard HDR. HDR+ easily deals with the bright, changing packaging reflections to create a more uniform image, making defects more visible. Since HDR+ does not require multiple images, it is a particularly valuable technology for high line speeds.
Even with HDR+, the combination of printed designs, reflective plastic, and visible contents behind the transparent parts of the plastic can create a visual field too complex for traditional machine vision.
Cognex AI-based technology brings the power of edge learning technology to the shop floor and makes it easy to automate challenging production tasks. The edge learning-based system trains on a small number of images, sometimes as few as two, to classify the inspected parts as acceptable and unacceptable, no matter how complex and ambiguous the visual field. Retraining to add a new content variant for component presence control and assembly verification is similarly easy and takes only a few seconds.
- Detect defects under challenging lighting conditions
- Reliably compensate for bright, changing reflections
- Ensure only satisfactorily packaged products reach the consumer