Studies on Authentication and Classification of Indian Herbal Medicinal Leaves using Image Processing Algorithms

Abstract

In biological sciences, images play a vital role in information data set. In ancient times people usually prefer herbal medicines for the treatment of ailment such as cough, sneezing, insect bite, cold and many more disease. In recent years due to the impact of Allopathic medicines usage of herbal leaves slowly vanished and nowadays the knowledge for herbal medicines amongst us is very less. This research paper focuses on the study of classification and authentication of herbal medicinal leaves based on automated leaf recognition system. newlineThe proposed system explains the techniques of image processing used for the classification of medicinal leaves. The dataset includes plethora of leaves under different species. The extracted features were classified under three categories such as color, shape and texture parameters comprising of 21 parameters for the classification. The test images were compared with the stored database images and the one with least dissimilarity was considered to be the closest match.The tests was carried out under different methods which includes, texture analysis, color moment classification and training the algorithm in neural network and real time implementation is done using Raspberry Pi processor. The accuracy achieved was newline99.2 % for the classification of herbal medicinal leaves. newlineLeaves such as NEEM, Hibiscus, Curry, Henna, Aloevera, Tulsi etc were captured using a high resolution camera of more than 16 pixels from various places such as Nursery, gardens, terrace and from other available platforms. The images are stored in the system for preprocessing and analyzing. A step by step process is followed for preprocessing such as converting and segmenting. For the analysis of the image Support Vector Machine (SVM) classifier is used which is the best suited classifier for morphological images. The dataset values are grouped for the leaves under same varieties and are used for classification.