Real time health monitoring method for tomato leaves

Tomatoes are one of the major horticulture crops consumed by majority of the population all over the world and it is very important to develop a real-time health monitoring of these crops for early detection of the diseases to minimize the risk of economic losses as well as an epidemic. In this thesis, we investigated image processing algorithms to monitor the health of the leaves whether tomatoes have diseases or not, categorizing them healthy or unhealthy leaves. Three different approaches were into considerations namely; Scale Invariant Feature Transform (SIFT), Color histogram and Principal Component Analysis (PCA), Gray Level Co-occurrence Matrix (GLCM) Texture features and Support Vector Machines (SVM). Images obtained from full color imaging sensors are utilized to train and test the algorithms. The algorithms are developed and executed by using MATLAB programming. The performance of three different approaches are compared and the simulations results indicated that GLCM algorithm outperforms the other approaches. We show that GLCM algorithm could be used as a real-time algorithm to monitor the health of the tomato crops and the approach could be extended and used for monitoring the health of the other crops in agriculture