DocumentCode :
2269555
Title :
Application of BP Neural Network to Sugarcane Diseased Spots Classification
Author :
Zhao, Jinhui ; Luo, Xiwen ; Liu, Muhua ; Yao, Mingyin
Author_Institution :
Coll. of Eng., Jiangxi Agric. Univ., Nanchang
Volume :
3
fYear :
2008
fDate :
20-22 Dec. 2008
Firstpage :
422
Lastpage :
425
Abstract :
Red rot disease and ring spot disease are two common diseases at the seedling stage of sugarcane. According to the image characteristics of diseased spots, sugarcane diseased spots classification using BP neural network is proposed. Firstly, the feature parameter combination of mean value of color component Cr, mean value of color component V and roundness is selected as the feature parameters of the following pattern recognition by orthogonal experiment design method. Then, BP neural network with 3 input neurons, 12 hidden neurons and 1 output neuron is constructed to distinguish between red rot diseased spots and ring spot diseased spots. The experimental results show that recognition accurate rates of rough set, fuzzy K near neighbor algorithm and BP neural network are 86%, 91% and 94%, respectively. BP neural network is more suitable to distinguish between the two diseased spots than fuzzy K near neighbor algorithm and rough set.
Keywords :
agriculture; fuzzy set theory; image classification; neural nets; rough set theory; BP neural network; fuzzy K near neighbor algorithm; image characteristics; pattern recognition; red rot disease; ring spot disease; rough set; seedling stage; sugarcane diseased spots classification; Artificial neural networks; Chromium; Color; Design methodology; Diseases; Fuzzy neural networks; Fuzzy sets; Neural networks; Neurons; Pattern recognition; BP Neural Network; Classification; Diseased Spots;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Intelligent Information Technology Application, 2008. IITA '08. Second International Symposium on
Conference_Location :
Shanghai
Print_ISBN :
978-0-7695-3497-8
Type :
conf
DOI :
10.1109/IITA.2008.447
Filename :
4740031
Link To Document :
بازگشت