On-line sorting maturity of Cherry Tomato by Machine Vision

Abstract

The cherry tomatoes online sorting according to their maturity is an important procedure after harvest. This research proposed an automated cherry tomato grading system base on machine vision. Three images of different angles are obtained from each cherry tomato, allowing the inspection of approximately 90% of the fruit surface. 9 features were extracted from the one cherry tomato images. In order to distinguish into three grades (immature, half ripe and ripe), Principal component analysis (PCA) and linear discrimination analysis (LDA) were used to analyze the features. The PCA results show that ripe cherry tomatoes are distinguished from immature and half ripe ones. 414 cherry tomatoes were tested by the online sorting system. The overall accuracy was up to 94.9%. Furthermore, the grading speed of the sorting line reaches 7 cherry tomatoes per second which meet the actual demand of many farms.