期刊名称：International Journal of Advances in Engineering and Management
电子版ISSN：2395-5252
出版年度：2020
卷号：2
期号：3
页码：104-108
DOI：10.35629/5252-02039195
语种：English
出版社：IJAEM JOURNAL
摘要：Here we propose animal identification system that employs image processing techniques. Firstly, the collected footprint images are pre-processed. Images are converted into gray-scale and boundaries of image are determined using canny algorithm. Further, footprint images are segmented. Gabour filter are used to extract features of segmented image. After feature extraction, features are reduced based on unsupervised model, we have used Principle Component Analysis (PCA) to reduce dimensionality features. Then reduced feature vectors are inputted into the classification model. Probabilistic Neural Network (PNN) is used for classification and identifying the animal class. Footprint datasets of 120 images are collected consisting of 10 different animal categories of variable training sets. Proposed system provides 93% of accuracy in detecting animal class.
关键词：Gabour filter;canny edge detector;PCA based dimensionality reduction and Pnn
