Correction of Leaf Area Overlap of Grafted Cucumber Plug Tray Seedlings Using Multispectral Imaging System

Leaf area estimation using multispectral imaging in grafted cucumber seedlings is often underestimated due to leaf overlap at later growth stages. This study investigates the use of multispectral imaging technology to estimate leaf area in grafted cucumber seedlings and proposes a method to improve estimation accuracy by introducing “days after grafting” (DAG) as a correction variable. For the experiments, the scion varieties ‘Goodmorning Backdadagi’, ‘NakwonSeongCheongJang’, and ‘Sinsedae’ were grafted onto the same rootstock ‘Heukjong’ (Cucurbita ficifolia), and images were acquired at 7, 14, and 21 days after grafting. The results show that including DAG as a correction variable significantly enhances the accuracy of image-based leaf area estimation, particularly in plug tray units, where R2 increased from 0.89 to 0.96 for ‘Goodmorning Backdadagi’, by effectively reducing errors caused by leaf overlap. Across all three varieties and both seasons (spring and summer), models incorporating DAG consistently showed higher accuracy in leaf area estimation than models without DAG. These results suggest that the method’s broad applicability is validated through comparisons across different seasons and varieties. Overall, this study provides a practical and accurate method for correcting leaf area estimation, with strong potential for application, particularly in seedling production and cultivation management.