In the meat industry, new methods are being invented regularly to deal with the conflict between increasing production volume and quality control (QC). Vision systems in particular are on the cutting edge of quality control measures.
A vision system can do any number of things, but generally, they use a camera to take images of a product and make a QC decision based on a physical trait. Because they are based entirely on images, they can run into problems with lighting and color.
To demonstrate these problems, we could consider the example of grading a beef carcass based on the quality of the ribeye cut. The ribeye is used to grade the whole carcass by industry and USDA standards, so this is an important process.
There are a number of different properties to look for when grading the ribeye. Fat content/marbling is an important factor in choosing whether the meat can be considered prime, choice, or select. For certain cattle, like the renowned Kobe, this is especially important for getting top dollar.
Fat detection can be harder than one would think, as it takes precise lighting, camera angle, and camera quality. The system needs to be able to pick it up and try to match it with a database of accepted images to see which image most closely resembles it and what grade that image ultimately was assigned. This image computation happens very quickly and can be automatic.
If the lighting in the area is dim or in a certain hue, the color contrast between meat and fat can be dulled, making it difficult for a system to pick up. To solve this, some places use ultra-violet (UV) light because it makes the color white pop out in great contrast. However, other colors may be harder to determine when UV is used. The right type of lighting depends largely on the tests needing to be made.
If a test for color is required, you must use some sort of light that is very color neutral and provides high contrast. With the ribeye, this can estimate the redness of the meat to assure that the carcass is healthy and fresh. The only trouble is that color can be deceiving in certain circumstances. If a beef carcass hasn’t been exposed to the air long enough after being opened up, its meat will have a much darker hue which could be misleading to the vision system if the carcasses passed through QC too early.
Installing a vision system takes very careful and precise planning, but when all of the bugs have been worked out, it can be an invaluable asset to any plant.
When it comes to automatic box labelingsystems, the use of a bar
We use two separate bar code scanners to do two different jobs. The first is referred to as a “pre-identification” scanner, and the second is called a “verification” scanner.
The function of a pre-identification scanner is to scan a label that has been previously applied to a box or pre-printed on a box to let the system controller know what product is contained in the box.
The verification scanner’s job is to scan the label that the system printed and applied to the box to be sure that it is a good, readable label.
With today’s technological advances, barcode scanners are being replaced with vision-based sensors. These sensors essentially work by taking a picture of the label, and by using the processing power of the sensor itself, it is able to read the barcode on the label and send that information to the system controller.
This type of sensor offers many advantages that were not available with the laser-based scanners:
- Ability to read both 1D and 2D codes
- Ability to store every scanned image to a file on the controller
- Ability to view images from several different sensors in real time
- Ability to grade the quality of each image, which allows the controller to signal preventative maintenance
- Better ability to read damaged codes
- Allows the use of multiple filter settings
This lets the sensor look at the image in a different way. For example, if the sensor is going to have 10 chances to read the barcode, you can set up the sensor to try 8 times with a setting that works best under normal circumstances, and if it fails to get a good read, it can look at the label a different way for the last 2 chances.
With these advances in technology, this type of sensor offers a lot of added functionality over standard barcode scanners for essentially the same cost.