Automatic Detection of Diseases in Leaves of Medicinal Plants Using Modified Logistic Regression Algorithm

• Published in: Wireless personal communications Vol. 131; no. 4; pp. 2573 - 2597
• Main Author: Meenakshi, T.
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.08.2023; Springer Nature B.V
• Subjects: Adaptive systems; Algorithms; Classification; Clustering; Communications Engineering; Computer Communication Networks; Cosmetics; Crop diseases; Damage; Deep learning; Diagnosis; Engineering; Feature extraction; Herbal medicine; Image enhancement; Image processing; Machine learning; Medical imaging; Networks; Pathogenesis; Performance evaluation; Plant diseases; Regression; Regression analysis; Signal,Image and Speech Processing; Adaptive gamma correction; Machine learning models; Disease detection; K-mean clustering logistic regression; GLCM; Classification
• ISSN: 0929-6212; 1572-834X
• DOI: 10.1007/s11277-023-10555-5

Medicinal plants are now more widely used by people than ever before, and are used in food products, cosmetics, religious deities, and especially in medical treatments. In agriculture now farmers are cultivating a wide range of crops related to medicinal plants. More crop damage is being done to farmers due to lack of awareness on the pathogenesis of medicinal plants, changes in nature, and lack of proper research on diseases of medicinal plants. Testing each plant by itself and making a diagnosis can be a costly endeavor for the farmer and requires more manpower. The impact of crop diseases on crop quality and yields is huge. If diseases can be identified and prevented at an initial stage, the yield can be protected from huge damage. Automatic diagnosis of medicinal plant leaf using image processing, deep learning and machine learning methods is increasingly underway. This work proposes a new variant method based on Modified Logistic Regression machine learning algorithm for identification of diseases in leaves of medicinal plants. In the proposed system adaptive gamma correction for enhancing features in the captured leaf image, k-mean clustering for segmenting healthy and diseased areas and GLCM and Logistic regression used for feature extraction & classification. Dataset has partitioned into different ratios, and its performance evaluation testing classification accuracy is compared between the proposed method and the existing logistic regression method, proposed method is showing better performance.