Image-based maturity detection for selective cauliflower harvesting in field condition

• A cauliflower curd maturity detection unit was developed for a selective cauliflower harvester. • Performance evaluation of different image processing algorithms was carried out. • The chromatic aberration method was found to be the most accurate among other algorithms. • The developed camera height compensation system performed very well with 94.0-97.0% accuracy. The traditional manual cauliflower harvesting is common worldwide, constituting up to 50% of production expenses due to its labor-intensive nature. Furthermore, since all cauliflower fruits do not mature simultaneously, multiple harvests are required from a single field, presenting a challenge for conventional cauliflower harvesting machines. Hence, there is a need for a selective cauliflower harvester capable of harvesting based on maturity levels. The present study developed a cauliflower curd maturity detection system for a selective cauliflower harvester. The present study used curd diameter as an indication of cauliflower maturity. The developed detection system uses an RGB camera to capture images, which are then further processed to determine the diameter of the cauliflower. Three algorithms—the color difference method, the color component ratio method, and the chromatic aberration method—were developed to accurately segment cauliflower. To achieve optimal segmentation performance with the developed algorithms, threshold values were optimized using the response surface methodology technique. When the chromatic aberration method was used at 42000 lux and a CA value of 232, it had the best segmentation performance (70.83% recall, 99.37% precision, and 99.08% accuracy). A camera height correction unit utilizing an ultrasonic sensor was incorporated into the system to enhance measurement precision by mitigating fluctuations in camera elevation in the field. The designed maturity-detecting unit effectively divides cauliflower curds, guaranteeing precise dimension assessment while accommodating changes in camera height. Harvesters can efficiently incorporate this created unit to selectively harvest cauliflower based on maturity stages.