Machine vision techniques for the evaluation of seedling quality based on leaf area

Abstract

Highly efficient automated transplanters in greenhouses are of great convenience to growers. These tools perform various tasks, including the removal of bad plugs and the fixing of empty cells in plug trays. Leaf area of a seedling is an important indicator of its quality. Here, a vision system was used to measure the leaf area in each cell to distinguish “bad” and “good” plugs. Based on the principle of proportion in area, the procedures for processing top-view seedling images and a method for calculating each the leaf area of each seedling in the plug tray were investigated. Overlapping of the leaves across the surface of the cell resulted in failures in identification, which is a key point to be resolved. A decision method combining the region centre of cross-border leaves, and a methodology for the improved watershed segmentation for overlapping leaf (OL) images, were developed. Seedlings of tomato, cucumber, aubergine and pepper, at suitable transplanting stages, were used to test the efficacy of the quality evaluation program. Through the segmentation of 40 seedling images (10 for each vegetable seedling), the improved watershed segmentation lessened the initial partitions by 45–55% compared with the conventional watershed algorithm. The OLs were successfully segmented. The relative identification accuracy of seedling quality was 98.6%, 96.4%, 98.6% and 95.2% for tomato, cucumber, aubergine and pepper, respectively. The errors were mainly attributed to horticultural practices. The results showed that this system of identifying seedling quality was suitable for application in automated transplanters.