Fungus image identification using K-Nearest Neighbor

2nd Nommensen International Conference on Technology and Engineering

Rahmat, R.F.^1 ; Aruan, T.^1 ; Purnamawati, S.^1 ; Faza, S.^1 ; Lini, T.Z.^1 ; Onrizal^2

Department of Information Technology, Faculty of Computer Science and Information Technology, Universitas Sumatera Utara, Medan, Indonesia^1

Faculty of Forestry, Universitas Sumatera Utara, Medan, Indonesia^2

关键词: Accuracy rate;    Conventional methods;    Identification process;    Image identification;    Image preprocessing;    K-nearest neighbors;    Living things;    Zoning method;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/420/1/012097/pdf
DOI  :  10.1088/1757-899X/420/1/012097

来源: IOP

【 摘 要 】

A fungus is a type of plants that have no chlorophyll, and its life depends on other living things. The fungus has various types. Some mushrooms are useful and utilized in the field of culinary, health, economy and agriculture, but there are poisonous ones that can cause infection and even death to humans. Research on fungus identification is insufficient, and the conventional method still becomes the main choice in identifying them. Therefore, an alternative is required to identify the poisonous mushrooms. The method proposed in this study is K-Nearest Neighbor. Mushroom image served as the input image of this image processing process. Before identification process, the image will go through image pre-processing step using gray-scaling, image segmentation using canny edge and thresholding, and feature extraction process using zoning method. 40 images of fungus plants were used to test the system. The accuracy rate of the system in poisonous mushroom identification is 90%.