Grading machines to inspect agricultural products by imaging technologies have been increasingly introduced recently in Japan. In fruit grading operations, it is one of the most important functions to inspect all sides of an object. In this study, two experiments on grading, based on full surface inspection of tomato fruit, were conducted by using a roller pin conveyor with a turning-over device and six image acquisition systems: 1) an experiment to turn over a fruit and 2) an experiment to synthetically grade a fruit based on each of the 6 images.
In the turning-over experiment, 100 tomatoes, which placed a roller pin conveyor with their calyxes down, were spun into upside-down postures through mechanically controlled roller pins. The results showed that this mechanism could be used for full fruit surface inspection, because about 80% success rate was obtained, whereas the success rate of smaller size fruits was lower than that of larger-size fruits due to their shape. After turning over, scattered spin angle of fruit showed variation for defect measurement, therefore some compensation of the result depending on the spin angle and improvement of spinning accuracy were desirable.
The results of the grading experiment confirmed that a set of the roller pins to turn fruit and six color TV cameras could inspect all sides of the fruit and that the grading results could roughly follow human operators' results, during the turning-over process of tomato fruit. However, grading results of fruits with cracks around calyxes different from human grades because of similar crack colors and thin calyx ends.To improve the grading results, it was observed that higher success rate of fruit turning over and higher resolution imaging technologies were required.
View full abstract