ISHS Acta Horticulturae 517: XXV International Horticultural Congress, Part 7: Quality of Horticultural Products
SIMULATION OF VISUAL QUALITY INSPECTION OF APPLES BY IMAGE ANALYSIS |
| Authors: | I. Paulus, E. Schrevens |
| Keywords: | image analysis, quality classification, standardisation, visual inspection tasks |
| Abstract:
Quality standards in horticulture distinguish high quality from less quality products. The meaning of this standardisation is to facilitate purchase of the better classified fruits and to give useful information to the consumers. In practise quality norms are defined in order to make the quality classification consistent. Nevertheless there exists a considerable disagreement between inspectors controlling the same products. To get a better understanding in human grading processes, it is important to get a better insight in the capability of quality staff to apply the current quality standards. The general objective of this research is the development of a methodology to compare the visual inspection of apples by human experts. In repeated experiments, quality assessors were asked to classify fruits applying the current quality standards. The apples were also analysed by an image processing system, extracting visual quality properties out of digitised video-images. Descriptors were developed to quantify colour, shape and size of the apples by processing four video images per apple. These objective measured features were connected with the human classification by binary decision trees. This resulted in a tool to compare the visual inspection of quality experts and to simulate the human quality classification. | |
Download Adobe Acrobat Reader (free software to read PDF files) | |
