Articles
PREDICTING THE MATURITY OF ORANGES WITH NON DESTRUCTIVE SENSORS
Article number
421_29
Pages
271 – 278
Language
Abstract
In the fragile market of fruit and vegetables, quality is an important factor.
In the packing stations, the fruits must be sorted out according to their quality.
This approach was applied to oranges for which various sensors were tested in order to predict the maturity.
Non-destructive sensors were used such as vision system, firmness sensor, near-infra red sensor, and reference measurement were made such as sugar content, acidity content, size and weight.
A panel of 3 experts provided 3 classes of fruits depending on their aspect.
A total of 320 oranges from the same variety (Salustiana) were tested with these sensors.
In the packing stations, the fruits must be sorted out according to their quality.
This approach was applied to oranges for which various sensors were tested in order to predict the maturity.
Non-destructive sensors were used such as vision system, firmness sensor, near-infra red sensor, and reference measurement were made such as sugar content, acidity content, size and weight.
A panel of 3 experts provided 3 classes of fruits depending on their aspect.
A total of 320 oranges from the same variety (Salustiana) were tested with these sensors.
Neural network techniques provided better results than multiple linear regression to predict maturity.
Vision system was very useful for the defect detection and color evaluation.
The combination of the various sensors remains a difficult task for the prediction of the maturity index.
One of the major constraint is the acidity level which can not be predicted with low-cost and non destructive sensors.
Future research will include the use of an artificial nose sensor that could detect rotten fruits.
Authors
V. Steinmetz, E. Biavati, E. Molto, R. Pons, I. Fornes
Keywords
Online Articles (31)