The current quality inspection of tomatoes on the vine is done by simple visual inspection of workers and quality controllers. Under the pressure of production, human errors and loss of attention, it is difficult to assure an objective and constant quality. Camera techniques are well able to deliver reproducible and consistent results and can scan all the products.
Assured quality camera techniqueThe first phase of the project is a feasibility study. The research should bring answers to the question whether, and with what camera technique it is possible to achieve a better sorted product than the sorting currently done by manual inspection done by the employees. Several product properties such as defects of the fruits (spots, blossom end rot, cracks, splits, etc.), defects of the stem and crown (yellow discolouration, mildew, mould and rot) and detection of the fruit maturity (green-red colouration) are examined.
Image analysis tomatoesRecently image data was collected using a number of camera different camera techniques including colour, hyper-spectral in the visible range (400nm-1000nm) and infrared range (900nm-1700nm) and chlorophyll fluorescence. An initial analysis of the data has shown that it should be possible to develop classification algorithms for many of the defects. A particularly interesting range of wavelengths for the detection of mould and fungi turned out to be the (near) infra-red region which cannot be observed with the human eye.
Currently, a lab setup is realized in Wageningen to image at a larger batch of tomatoes on the vine using the selected camera setup. Then, computer vision algorithms for real-time analysis will be developed. If sufficiently high classification scores can be reached an inspection module will be realized. This module will be designed such that it can be integrated in the packaging line.