Quality Assessment and Classification of Basil Using Computer Vision

Abstract

The natural products are inexpensive, non-toxic, and have fewer side effects. Thus, their demand especially herbs based medical products, health products, nutritional supplements, cosmetics etc. are increasing worldwide. Majority of medicines are ready from medicinal plants. But, due to diseases medicinal plants growth diminish severely. The diseases possess threats to economic, and production status in the medicinal industry worldwide. So, it is mandatory to continuous measuring quality of plants to predict the disease extremity. Earlier, manual observation is used to analyze quality but somehow it is tedious, inconsistent and costly. By now, studies show that digital image processing methods work as effective tools for the identification and classification of plants diseases. Medicinal plants as basil, neem, aloe, pepper, and turmeric are widely used for preparation of Ayurvedic and allopathic medicines. Particularly, basil has an intense significance in medicine prospective. So, basil disease detection and classification using computer vision is the motivation of presented work. Pathologists focus on diseases in different parts of the plant like roots, kernel, stem and leave. The presented thesis concentrate, particularly on leaves. The work present in this thesis is focus on to design a new framework for segmentation, feature extraction and classification. After, data set preparation a new segmentation technique with neutrosophic logic is used to detect and identify region of disease. Features are extracted from segmented regions using amalgamation of texture and color features. New texture feature is also introduced named as bin binary pattern. Then, we used different classification models for diseases predication. As comparison to existing segmentation techniques, proposed method gives promising results. A classification algorithm using survival of fittest approach is proposed in other work. Best solution is obtained through the fitness function with minimum distance and maximum similarity values using maximum aggregation analysis. As comparison to existing machine learning methods; proposed classification method provides best results. Moreover, present thesis contributes in the area of healthcare and medicines, which plays significant role in daily life.