Image Retrieval of Poisonous Mushrooms Based on Relevance Feedback and Clustering Algorithm

摘要

Poisonous mushrooms have no significant difference with the edible fungus. In this paper, image retrieval technology based on color, shape and texture characteristic are applied to identify poisonous mushrooms With relevance feedback mechanism and clustering algorithm used to improve the retrieval efficiency. Clustering can divide similar images into a same class and greatly reduce the range of image retrieval. From the retrieved image, representative clustering images are selected to the users who can choose a satisfactory clustering from them and provide feedback to the system. Experimental results with this method show effectiveness of identifying unknown poisonous mushrooms.