Application Research of Automatic Rice Pest Identification based on SOFM
Aimed at the problem of the automatic detection and classification identification for rice pests, an automatic rice pest recognition system based on skeleton morphology feature is designed. Four skeleton morphology features such as skeleton point branch number, skeleton accumulative total length, Xlength/Y-length ratio and circularity are automatically extracted from rice pest image, and loaded into a neural network SOFM as input factors. The neural network is trained with extracted features from 20 images of rice pests. In this way the problem of rice pest recognition is solved better. This system is simple, reliable, economic, expandability and has a great agricultural application value.
skeleton morphology feature extraction image recognition automatic identification neural network SOFM
Pengfei Wu Yi Wang
Dept. of Computer Science College of Basic sciences, Huazhong Agricultural University Wuhan, China
国际会议
太原
英文
640-643
2011-02-26（万方平台首次上网日期，不代表论文的发表时间）