What is feature matching?
Feature matching is one of the technologies used in computer vision. With this technology an algorithm searches for unique sets of pixels on an image and compares them to other images in a database. Eventually the software can match two images with each other to select the one that is most similar.
The technology
With feature matching technology a program can extract a number of unique pixel features from images in a database. This database can for instance be labels that have to be printed or images that have to be put on custom gifts. These features are unique groups of pixels that occur in an image, like edges, corners of specific color clusters. If the actual product then passes a camera, the software finds the features in a similar way and matches them to the database. The database article that is most similar to the one on the camera is then selected. By connecting to other systems it is possible to retrieve more information, to print stickers for the packaging or send information to ERP systems for instance.
Applications
Product labels
A camera can identify specific labels in boxes and finds the corresponding order number in a database. It connects to a printer that labels the box accordingly and sends the information to an ERP system.
Personal photo mugs
Hundreds of personalized photo mugs go into an oven and when they come out no one knows which order they belong to. A vision system scans each mug and retrieves the order number and prints a label.
Branches
- Printing industry
- Custom gift production
Pro's and cons of feature matching
Pro's
- Prevents errors in order matching by specifically recognizing the image.
- No manual control needed for each order.
- Saves a lot of time for employees to manually find the right order.
- As easy as scanning a barcode.
Cons
- You need a production flow where your orders can be pre processed before passing the camera.
