In this research we built a working model of a date fruit grading and sorting system including both: the hardware and the
software. The working prototype of the system is shown in
Fig. 9below. The hardware includes the conveyer, camera control and helm control systems. The software system analyzes
the fruit image and classifies them. The maximum accuracy
of the system is 80% which is attained by model 2 in classifying
the grade 2 fruit.
We observed problems in detecting the flabbiness from the
color. An impact sensor might improve flabbiness detection.
Our fruit quality grading into three grades was based on human perception. A formal feature distribution based method
need to be developed to determine the fruit quality grade from
the samples. We feel that this should improve the classification
accuracy. To determine the feature based grades we are investigating the suitability of the unsupervised learning techniques.
We are in the process of applying self organizing map to obtain
the fruit grade clusters using the feature distribution in large
samples