One of Romania’s leading ice cream producers implemented an automated vision solution that can detect badly shaped, broken or stitched-together ice creams and facilitate their fast removal from the production line.
With more than 10 years' experience in industrial automation, PHILRO Industrial of Voluntari (Romania) develops and produces machine-vision systems for any industry where such solutions are needed.
Founded in 1994, Betty Ice of Suceava (Romania) is currently the leading Romanian ice cream producer. With state-of-the-art equipment and a motivated team, Betty Ice has succeeded in being one of the modern ice cream factories in Europe. During the production process, ice creams may break or stick together. Visual inspection by human operators is not infallible, and can result in inappropriate products delivered to consumers.
The Romanian food and beverage market has become extremely competitive and consumers expect high quality on a consistent basis. Those who buy a pack of ice cream expect that size, weight and quality of each ice cream to be the same.
Achieving consistent quality for each delivered product in the most effective way and without any disturbances in the production process became an important objective for Betty Ice.
Ice cream on the assembly line
Before this automated detection system, human operators were supervising the final part of the ice cream production line, visually detecting defective products and chasing them off the line before entering the packaging stage.
To enhance the capability of their quality control process, Betty Ice contacted PHILRO Industrial‘s Machine Vision Division, in search of a solution. The result is a vision-based inspection system that can detect badly shaped, broken or stitched together ice creams, and facilitates their fast removal from the production line.
PHILRO Industrial installed a quality control system based on machine-vision at the Betty Ice ice cream factory, using professional cameras, dedicated hardware and software based on MVTec's HALCON libraries, to visualize production flow and detect defective ice creams (broken, glued, with different shapes, etc.). All hardware components are embedded and connected with the operator’s computer through an Ethernet network.
The system consists of the following components:
- 4 Basler Ace aca1300-30gm video cameras with 12 mm lenses, waterproof enclosures and GigEthernet communication;
- Adam 6052 for process control;
- NUC6i5SYH computer for image processing;
- Sick inductive sensor, used as a trigger for taking pictures.
On the production line, ice creams are hanging on parallel supportive rows, 16 ice creams per row. When each row with ice creams reaches the SICK sensor, the 4 cameras are triggered for the image capturing procedure. After the images are taken, the analyzing process begins. For every row with ice creams, 4 images (4 cameras for 16 ice creams) are taken. To shorten analyzing time, each image is processed on a separate execution thread.
HALCON imaging software is used in the customized software application, which analyzes the image to ascertain if every ice cream is the correct size.
The system is using several HALCON operators:
- gen_ellipse: Creating a region of interest for each ice cream, as they always came in the same position;
- reduce_domain: Restrict the image domain to analyze each ice cream separately;
- threshold: Used to suppress the background;
- connection: Used to create a single entity, namely ice cream, from many pixels which define the image;
- fill_up: It is assumed that an ice cream contour may have inside pixels that have not been highlighted, so the ice contour is filled up;
- area_center: Measure the area of the created object (ice cream) and check if it has a certain size, which means the area is OK in terms of quality.
The system is able to check up to 1,000 ice creams per minute. In the front end of the software application, the operator can select from a drop down menu for product changeover, and receives continuous feedback from the detection system about the production process. If some ice creams are detected as defective, an alarm is triggered for a real time removal from the production line.
The application stores all gathered data in a database, which may be connected with company’s ERP for statistical purposes and enhanced traceability. Beside significant savings achieved by automating the quality control process, the solution implemented by PHILRO Industrial at Betty Ice ice cream factory leads to a significant reduction in the defective products rate, which will increase Betty Ice’s customers’ satisfaction and brand reputation.
Images at right show ice cream images before processing (top) and after (bottom).
Written by Mihai Nasta and MÄdÄlin Roventa of Philro Industrial.