Self-service scales use imaging to identify fruit and veg
Mettler Toledo has developed supermarket fruit and vegetable scales that incorporate an industrial camera. The system is designed to help shoppers choose the correct variety of fruit and veg from the menu of self-service scales.
Mettler Toledo’s engineers integrated a uEye USB camera from IDS Imaging Development Systems in its PC-based UC3-GTT-P scales. Measuring only 3.5cm, the board-level camera is concealed inside a curved metal arm mounted to the scale's screen. The machine vision system has been developed for the shoppers' convenience and makes the self-service scales much easier to use.
The imaging and analysis takes less than a second and four possible matches are presented to the consumer for selection. Due to various disturbance factors a fully automatic system cannot be implemented. Highly fluctuating light conditions or covered areas in the image affect the analysis. If conditions are good, however, the camera system achieves high accuracy even if the weighed products are in plastic bags.
The image analysis is based on colour imaging. The colours detected in the produce are compared with a stored list of colours that are assigned to individual types of fruits and vegetables. The system analyses not just single colours, but also colour combinations. Therefore a specific combination of red, green and yellow could be recognised as mixed peppers, for example.
A combined analysis approach helps distinguish between products of a similar colour, such as lemons and bananas. Characteristics like shape and texture are additionally evaluated. Mettler Toledo has also developed algorithms in cooperation with the Fraunhofer Institute for Information and Data Processing to combat imaging through plastic bags, which could otherwise complicate the analysis.
The camera size, its USB 2.0 connection and its colour quality were key considerations when selecting the IDS camera. Walter Grom, the engineer in charge of the project, stated: ‘An important factor in our decision was of course an excellent image quality, which the uEye provides.’ Grom added that the Linux driver was another reason for choosing the camera.
The UI-1226LE-C, which captures up to 87fps at the full 752 x 480 pixel resolution, was used. Another advantage is that the uEye's automatic white balance works reliably even against the blue background of the scale.
In Germany, the scales are installed in more than 300 food retailers to date. Other European countries have also started using the system and are planning to expand its use.