Detection of Red Tomato on Plants using Image Processing Techniques

Abstract

Tomatoes are the best-known grown fruit in greenhouses that have been recently attempted to be picked up automatically. Tomato is a plant which its fruit does not ripe simultaneously, therefore it is necessary to develop an algorithm to distinguish red tomatoes. In the current study, a new segmentation algorithm based on region growing was proposed for guiding a robot to pick up red tomatoes. For this purpose, several colour images of tomato plants were acquired in a greenhouse. The colour images of tomato were captured under natural light, without any artificial lighting equipment. To recognize red tomatoes form non-red ones, at first background of images were removed. For removing the background, subtraction of red and green components (R-G) was applied. Usually tomatoes touch together, so separating touching tomatoes was next step. In this step, the watershed algorithm was used that was followed by improving process. Afterwards, red tomato was detected by the region growing approach. Results obtained from testing the developed algorithm showed an encouraging accuracy (82.38%) to develop an expert system for online recognition of red tomatoes.