DocumentCode :
1720454
Title :
Classification of flotation froth based on CCM texture feature
Author :
Chen Ning ; Zhang Lei ; Gui Weihua ; Yang Chunhua
Author_Institution :
Sch. of Inf. Sci. & Eng., Central South Univ., Changsha, China
fYear :
2013
Firstpage :
4735
Lastpage :
4740
Abstract :
The surface texture of flotation froth is closely related to flotation production conditions. In order to conduct accurate classification and recognition to flotation production conditions, a method of BP neural network classification based on CCM texture feature extraction is proposed. First, the feature statistics which are extracted to characterize the froth texture condition are treated as texture feature parameters. Then, a principal component analysis is used to extract two characteristics which contain the main contents of information, COR and ASM. Finally, according to the peculiarity of the classification data, the standard BP neural network with 140 nodes in input layer, 14 nodes in hidden layer and 4 nodes in output layer, is used to classify the froth flotation state. Experimental results are given to demonstrate that the method can accurately classify the froth flotation state and obtain better clustering quality of flotation froth. It will provide guides for improving the operation of flotation.
Keywords :
backpropagation; feature extraction; flotation (process); image classification; image texture; neural nets; pattern clustering; principal component analysis; production engineering computing; ASM; BP neural network classification; CCM texture feature extraction; COR; clustering quality; feature statistics; flotation froth classification; flotation production conditions; principal component analysis; surface texture; Educational institutions; Feature extraction; Image color analysis; Manufacturing processes; Neural networks; Principal component analysis; BP artificial neural networks; color co-occurrence matrix; flotation production condition; froth image; principal component analysis;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Control Conference (CCC), 2013 32nd Chinese
Conference_Location :
Xi´an
Type :
conf
Filename :
6640257
Link To Document :
بازگشت