Abstract

The export value of flower production is very high in many countries. Among these export flowers, orchids are considered one of the most valuable flowers. In order to increase orchid production, it is important to take closely care of orchid in the cultivation period. Due to the plant diseases may cause the economic losses in agricultural industry, the daily inspection and early recognition of plant diseases are necessary. Detection and prevention of plant diseases are a worldwide agricultural problem. Many researchers have proposed biocontrol or IT technology to handle plant disease problems. Some researchers applied image recognition to find out leaf problems such as banana leaves, alfalfa leaves, and citrus leaves. However, these researches all focus on trees or fruits in the lab. They did not provide the long-distance leaf identification for orchid flowers. To help flower farmers and to enhance the quality of production, this research has proposed a machine learning method to capture the image of orchid leaf and to identify the leaf disease. This research analyzes the leaf image with feature space, i.e., HSI, RGB, and grayscale. We use histogram to analysis the leaf color and we analyze the green color threshold so that the image can be classified into various color zones. Based on the generated threshold, this research is able to segment the leaf image into healthy area and unhealthy area. Finally, we use the Artificial Neural Network (ANN) and Deep Learning ANN to learn the image patterns of orchid leaf. Our proposed method is then applied to identify the orchid leaves and to determine whether the orchid is healthy or sick. With our proposed model, the accuracy of recognizing the leaf disease can achieve 100% for training data and 90% for testing data. This research enables flower farmers to recognize the orchid disease and can prevent the disease in early stage. As a result, the farmer can take better care to the orchid plants and enhance the cultivation of orchid production.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call