Quantitation Of Wood Texture By Digital Image Processing

Texture is one of the most crucial natural attributes of wood, and is also the important foundation of appraising and utilizing wood, which has been regarded as central content in the research of wood physics and wooden environmentology. The quantitation of wood texture has certain scientific and practical applied value, but it has also puzzled the wood educational circles for long time. Introduction of digital image processing technology would offer double feasibilities on theory and practice for solving this difficult problem.This text first analyzed texture features of wood from psychological physics, and defined physical quantifiers on embodying the texture characteristics of wood; Then it advanced the pertinent detecting method on the physical quantifiers; Also some well-known textural algorithms and models that already approved were carried on to analyze the spatial exhibition, frequency spectrum characteristic; At last, it integrated the characteristics and excellencies of every algorithm together, and extracted a series of effective texture parameters, and set up wood texture quantitation system.The study has maken progresses at following aspects:(1) It analyzed the textural characteristics of wood from psychological physics attribute, and summed up a set of physical words which describes wood textural characteristics from human vision and image processing for the first time, which including: Textural hue, Textural conscious luminance (namely is luminance studied in the chromatology or gray level studied in digital image processing), Textural orientation, Textural periodicity, Textural shapes, Textural width, Textural spacing, Textural roughness, Textural consistency, Textural density, Textural intensity, Textural gradient, Textural contrast, Complexity of texture, Regularization of texture, etc..(2) It launched the researches on detection of pertinent textural characteristics of wood, and realized: ① In textural hue, measuring the chromatic indexes in RGB chroma system, XYZ chroma system, L*a*b* chroma system, HSI chroma system, and Munsell chroma system, and choosing HSI chroma system to express wood textural hue; ② In textural conscious luminance, analyzing the gray histogram characteristics of the whole texture and textural hues of wood, and denoting them with quantitative statistical parameters; ③ In textural shapes and textural orientation, filtrating the optimal functions of BWPERIM and BWMORPH in detecting wood textural shapes, and executing Radon transform to extract wood textural angles automatically; ④ In textural periodicity, textural width and spacing, inspecting the periodical alteration of wood textural grays, as well as their self-similarity characteristic, then by transversal linear scanning on pixels, having finished the measuration to textural intensity, textural gradient, textural width, texture spacing, and textural periods width.(3) It applied Spatial autocorrelation function, Spatial gray level cooccurrence matrix,Gray level run-length matrix to analyze the spatial second combination mode of textural paired pixels, and received textural main components of signifying the degree of textural homogeneity and consistency, textural intensity and contrast, textural density, textural brightness, and textural run-length and gray distribution. In textural fractal, calculating the fractal dimensions of wood texture, as a token of textural roughness and complexity.(4) It applied Fast Fourier transform spectrum and Wavelet to analyze wood textural features from multiscale and multifrequency domain, extracted wavelet energy distributing parameters and wavelet energy distributing proportion parameters.(5) On the basis of analyzing and comparing every textural algorithm synthetically, it established the feature parameter system of wood texture, which including: ① Hue (H), ② Saturation (S), ③ Illuminance (I), ④ Contrast (CON), ⑤ Angular second moment (ASM), ⑥ Sum of variances (SV), ⑦ Long run emphasis (LRE), ⑧ Fractal dimension (FD), ⑨ Wavelet horizontal energy distributing proportion (EPLH).(6) It examined the application effects of textural feature parameter system in wood science research through experimental designs:① Application of textural feature parameters as inputting neural nodes in automatic classification of wood species; ② Application of textural feature parameters in wood textural contents searches based on maximal similarity theory; ③ Application of textural feature parameters in the simulation and reconstruction of wood texture; ④ Application of textural feature parameters in supervising wood processing technics; ⑤ Application of visual physical magnitudes improved by textural feature parameters in predicting the psycological evaluation and environmentological quality of wood. Consequently, by experimental results, it affirmed the capabilities of wood textural featural parameter system, and stated its potential utilizing values.