Design, Development and Evaluation of an Orange Sorter Based on Machine Vision and Artificial Neural Network Techniques

The high production of orange fruit in Iran calls for quality sorting of this product as a requirement for entering global markets. This study was devoted to the development of an automatic fruit sorter based on size. The hardware consisted of two units. An image acquisition apparatus equipped with a camera، a robotic arm and controller circuits. The second unit consisted of a robotic actuator with required electronic circuits. For sorting purposes، an appropriate image processing technique was applied and two models of size thresholds were developed and incorporated in a number of image processing algorithms، which were، in turn، combined with Artificial Neural Network (ANN) techniques for classifying purposes. Multi Layer Perceptron models with various training functions and diverse numbers of neurons were also applied. Each algorithm was used to sort oranges into desired size groups (Small، Medium and Large). The sorter test rig was able to classify the product into three categories with considerably low errors. Although all twelve algorithms had acceptable results، those based on Red and Green segmentation were more satisfactory. For real time evaluation purposes، four algorithms، segmenting based on R color band، and two size threshold models were combined to form 8 comprehensive algorithms، which were used along with the ANN model at the evaluation stage. Results showed that algorithms based on Area، Perimeter and the ANN model، exhibited lower errors. Sorting records of each algorithm were compared to the relevant sorting data brought about by experts. Results show that sorting error can be as low as 1. 1%. Although the average capacity of the single sorter was limited to 1 t. h-1، the capacity can be markedly increased by adapting a bank of sorters in parallel mode. The study revealed that orange fruits can be sorted using the introduced techniques at high speed، high accuracy and low costs.