A Novel Method for Leaf Area Estimation based on Hough Transform

Abstract

Manual methods of Leaf area estimation are very time consuming and tiresome processes and these are generally inaccurate and subjective. A new approach is presented in this paper to make this process more accurate and user friendly by using Digital Image Processing techniques on leaves of healthy and diseased plants. A coin of five rupees is taken as a reference object to estimate the leaf area. Coins have been used in previous approaches in literature but none of the proposed methods have segmented coins accurately. Our method proposes a new approach for coin segmentation by using Hough transform. The main advantage of our proposed method is the area estimation of not only healthy leaves but also the diseased ones. Our method has been tested on four kinds of plants i.e. Potato, Pigeon Pea, Green Gram and Black Gram. Twenty leaves have been taken for each of these four plants in which ten leaves were healthy and remaining ten were of the diseased plants. The results obtained from this method were compared to the graph paper method. The average error rate of leaf area estimation of our proposed method are 1.49%, 1.11% , 1.71% and 1.31% for potato, Pigeon Pea, Green Gram and Black Gram respectively as compared to the results of graph paper method. It proves that our proposed method is more accurate for leaf area estimation of healthy and diseased leaves.