A Feature-Based Machine Learning Framework for Plant Disease Recognition Using Color and Texture Analysis

Abstract

One of Tamil Nadu's staple crops, paddy guarantees a sufficient supply of food for millions of farmers and promotes socioeconomic prosperity in the Chidambaram area of the Cuddalore district. Nevertheless, it has been discovered to be extremely vulnerable to a number of diseases, including bacterial leaf blight, blast, and brown spot, which can significantly reduce yield and quality.Since the conventional methods of diagnosing these diseases are extremely time-consuming, laborious, and prone to human error, early and accurate detection are crucial for prompt intervention and treatment. In order to develop an effective, scalable, and efficient solution, this study proposes a revolutionary categorization of paddy disease utilizing image analysis and deep learning approaches based on cutting-edge developments in artificial intelligence. In this work, we introduced a Feature-Based Machine Learning Framework for Plant Disease Recognition Using Color and Texture Analysis (FBMLF-PDRCTA) Technique. First, a Sobel operator is used to segment the lesion's edge after the sample picture has been smoothed using median filtering and histogram equalization. This greatly lowers background information and enhances image quality. Next, using color and texture features, the image's matching feature parameters are retrieved. Finally, Support Vector Machine is one of the most commonly used machine learning methods for categorisation. According to the study findings, when the total amount of node in the concealed layer is changed to 98, the FBMLF-PDRCTA's detection rate may reach up to 95.8%. The recognition technique developed with an SVM, having good accuracy, may effectively address the drawbacks of manual categorisation using the colour and textural features of a rice sheath blight image.