Font Size: a A A

Research Of Wooden Plank Classification Method Based On Color Characteritics

Posted on:2011-07-03Degree:MasterType:Thesis
Country:ChinaCandidate:B B ZhaoFull Text:PDF
GTID:2178360308471201Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
Color is an important characteristic of woodiness board of nature,and people tend to judge the quality of plank according to the color.In the woodiness floor processing and woodiness furniture industry especially,the color of plank is an important index of evaluating product quality. In view of that there is no national standard and industry now to define and describe the wood surface color, to combine image processing and pattern recognition met-hod,the theory establishes a set of parameters of the system which could show woodiness board surface color characteristics.It will provide an important reference and theoretical basis for automatic classif-ication of woodiness board surface color.The study summarizes image classification methods based on.color characteristics firstly-,and introduces common color space models and representation methods of color characteris-ticcccccs.Then color histogram statistical characteristics, color entropy characteristics and do-minating colors are used color space model. Some common classifier theorys and design methods are declared in the paper,which are k-Nearest Neighbor(KNN),BP neural network,Probabilistic Neural Network(PNN) and support vector machine(SVM).The important of the research is three extracting methods of wooden plank color charact-eristics.The first is based on ascending wavelet transformation to extract color characteristics. Wood samples transformed from RGB color space model to HSV color space model,and then every image is separated into four parts to abstract entropy of three color components,H,S and V.Twelve characteristic parameters are formed,and three classifiers,-KNN,PNN and SVM,are used to classification and Simulation experiment.The second is a met-hod of weighted features color hist-ogram statistical characteristics.The method adjusts weight of the original eighteen characteris-tic parameters and chooses characters according to Genetic Algorithm(GA).Lastly two classify-ers,KNN and BP,are used to classification and Simulation experiment.The third research appro-ach is based on global color and local color,and forms six global colors and sixteen local color. Genetic Algorithm(GA) is used to optimize characters,and KNN,PNN and SVM are the class-ifiers to classification and Simulation experiment.Contrasting the three extracting methods,the system of woodiness board surface color ch-aracteristics is definited which contributes to the wood production process and provides the basis of the research of other image processings and classifications.
Keywords/Search Tags:Pattern recognition, Wooden plank color, The feature extraction, Feature selection, Classifier
PDF Full Text Request
Related items