Digital image processing method for estimating leaf length and width tested using kiwifruit leaves (Actinidia chinensis Planch).

Abstract

In this study, an image-analysis-based method is proposed to estimate the length and width of plant leaves, some of which have concave tips and petiole insertions. Here, kiwifruit leaves are used. Using the proposed method, orthophotographs of the leaves were obtained through image segmentation and image correction. The coordinates of the pixels along the edge of the kiwifruit leaf were then extracted. Finally, the leaf length was determined based on the sum of the number of pixels between the leaf tip and petiole insertion, and the leaf width was calculated based on the sum of the number of pixels for the widest region of the leaf. The experimental results were evaluated by analyzing the mean absolute error (MAE), root mean square error (RMSE), and accuracy. The maximum MAE and RMSE for the three examined cultivars are 0.281 and 0.366 cm, respectively. The analysis also showed that the proposed method outperforms other prevalent approaches, achieving high accuracy rates of 98.88% and 98.83% for the length and width of kiwifruit leaves, respectively. The low MAE and RMSE and high accuracy prove the capability of the proposed method with regard to calculating the length and width of such leaves. In addition, the proposed method can be extended to other plants whose leaf shape is comparable to that of a kiwifruit leaf.