An Automated System for Identification of the Medicinal Leaf using MKSVM

The primary source of traditional medicine is found in medicinal plants. And these protect human health. The resource preservation towards traditional medicine has important implications found by the R&D of medicine leaf. Identifying the medicinal plants manually is a time-consuming process that requires the help of experts for plant identification. This paper comes up with a robotic system for the classification in the medical field, which is towards restricting manual classification, which is based on medicinal plant identification. The proposed system has three modules, namely pre-processing of the image, image feature extraction, and later the image classification. In the initial pre-processing step, the conversion of RGB is conducted to extract the green band in the input images. The median filter method is used to remove noise present in the input images obtained from the green band. In the second step, after pre-processing, some of the features like shape, color, and texture, are extracted from the pre-processed image. The multi kernel-based support vector machine (MKSVM) classifier is used to classify the image as medicinal or regular leaf by the extracted features. The performance of the recommended methodology is examined in terms of different metrics, and performance is compared against different classification methods. Achived accuracy is 95.8%.