Image Processing for Diagnosis Rice Plant Diseases Using the Fuzzy System

• Published in: 2020 International Conference on Computer Science and Its Application in Agriculture (ICOSICA) pp. 1 - 5
• Main Authors: Rifa'i, Anwar, Mahdiana, Deni
• Format: Conference Proceeding
• Language: English
• Published: IEEE 16-09-2020
• Subjects: Diseases; Fuzzy system; Fuzzy systems; GUI; Microorganisms; Rice plant disease; Sugeno; Testing; Training data; User experience
• DOI: 10.1109/ICOSICA49951.2020.9243274

The biggest challenge for farmers in producing quality rice plants is rice plant disease. To reduce the impact caused by rice disease, early detection is needed. Farmers usually detect rice diseases based on the characteristics that appear in rice plants. Unfortunately, diagnoses made by farmers often vary due to differences in experience and knowledge. Rice plant disease diagnosis through the laboratory is considered less effective because it requires long time and a large amount of cost. This study aims to build a system for detecting rice diseases that can be used quickly with accurate results. The system was built using a fuzzy system with ten inputs. This input is the result of the extraction of rice plant images, i.e. contrast, correlation, energy, homogeneity, average, variance, kurtosis, entropy, standard deviation, and skewness. The fuzzification process is carried out using the Gauss membership function, the fuzzy inference is done using the Sugeno method, and the defuzzification process uses the weight average method. Output variables are classified into three sets, i.e. Bacterial leaf blight, Brown spot, and Leaf smut. Graphical user interface was used to improve the user experience in using fuzzy system. The accuracy of the detection system of rice disease produced was 94,792% based on training data and 91,667% based on testing data.