Font Size: a A A

The Research Of Wood Surface Texture Classification Based On MRF

Posted on:2007-04-12Degree:MasterType:Thesis
Country:ChinaCandidate:L ShiFull Text:PDF
GTID:2178360185955547Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
The wood surface texture is an important component of the wood visual characteristics, such natural pattern filled with aesthetic is complicated, and it is concerned with the sense of the woodwork and economic benefits directly. So it is important to signify and study.This subject is a technological unit of the natural science fund in Heilongjiang projects "The research of wood surface visual characteristics"(C2004-03). The preconditioning method of the wood surface texture image is introduced at first, and the image processing of wood surface texture is expounded based on experiments to deal with the images conveniently, such as sample, grey level transformation and noise elimination.Markov Random Field (MRP) is extension of Markov Processing at the two-dimentional parameters set. MRP is probability model to express texture gathering in mathematics. The statistical parameters can express the magnitude and direction of pixels neighborhood, and describe the random characters of image texture. So Markov Random Field model is built according to wood surface texture characteristic, and the parameters of wood surface texture is acquired after estimation.With artificial neural network (ANN), the wood texture is classified according to MRF characteristic parameters. The neural network pattern-recognition is a new study direction of pattern-recognition in recent years. The experimental result proves that MRF parameters is effective, the ANN pattern-recognition is feasible.The image matrix is decomposed to different matrix in different scale and direction with wavelet multi-analysis theory, each matrix represents message in 4 directions, such as approximate value, horizontal details, vertical details and diagonal details. Next, the characteristic parameters of MRF are acquired in wavelet decomposing regions, and are classified with ANN. The experimental result proves that the study of texture characteristic parameters after wavelet decomposing set up a new direction for the texture.
Keywords/Search Tags:wood surface texture, Markov Random Field(MRF), neural network(NN), classification
PDF Full Text Request
Related items