Shape Features based classification of Herbal Plants from its Powder using Microscopic Image

Abstract

An identification and classification of herbal plants from its powder using the microscopic image is a challenging task. In this paper, a new approach for identification and classification of Indian herbal plants liquorice, rhubarb and dhatura using the microscopic image is proposed. This paper evaluates the effectiveness of the shape based features with a different classifier for classification of herbal plants. The analysis of microscopic images performed in three stages: Segmentation, Feature Extraction and Classification. To detect the object from the microscopic images we have manually crop the object. For automatic detection of the object, we have applied the Extended Quantum Cut (EQCUT) method of segmentation. After object detection, we have computed the three shape features, including compactness, moments and Fourier descriptors for each object. For evaluating the effectiveness of the shape based features set, various combinations of the three shape features were investigated with support vector machine, K- nearest neighbor and ensemble classifier. Highest classification accuracy of 94.9 % achieved using bagged tree ensemble classifier when Combining all shape features.