CN105629724A

CN105629724A - Novel precious wood-simulated dying type computer color matching method

Info

Publication number: CN105629724A
Application number: CN201410583072.3A
Authority: CN
Inventors: 李滨; 郭明辉
Original assignee: Northeast Forestry University
Current assignee: Northeast Forestry University
Priority date: 2014-10-27
Filing date: 2014-10-27
Publication date: 2016-06-01

Abstract

The invention provides a novel precious wood-simulated dying type computer color matching method. According to the method, firstly, principal anatomical factors that influence the dying effects of both the coniferous timber and the broadleaf timber are determined, and the principal anatomical factors are adopted as the reference effect data of the computer color matching method. Secondly, a recipe prediction model is established based on an RBF neural network, and an RBF neural network model is optimized according to a hidden node-based method. The optimized model effectively solves the problem in the prior art that an original model is slow in acquaintance speed. The speed of the above optimized model almost meets the online training standard. Finally, a fuzzy neural network is introduced to the prediction model, and an improved subordinating degree function is provided based on the features of timbers. In this way, a fuzzy neural network prediction model based on the improved subordinating degree function is established. The above model is fast in running speed, and easy in parameter setting. Since no more setting condition is required, the time is greatly saved. Moreover, the method is small in calculation error, and the maximum error thereof is only 1.25%. Therefore, the obtained model is better in generalization ability.

Description

A kind of computer for colouring method of the imitative treasure's dyeing of novel timber

Technical field

The present invention relates to the computer for colouring method of the imitative treasure's dyeing of a kind of novel timber, belong to wood staining technology class.

Background technology

Wood color is one of key factor determining consumer's impression, in order to improve decoration function and the value of the product of woodwork, it is achieved the efficient utilization of artificial forest wood, it is necessary to by staining technique improvement material inferior. An important step in wood staining is matched colors exactly.

Adopting Man-made Color Matching, it is high to the competency profiling of color matching personnel, not only time-consuming but also be difficult in adapt to the requirement that modern industry produces, and cost is high, poor accuracy. When determining technique, the method that computer intelligence is matched colors is used in wood staining match color procedure, it is possible to accelerate the speed that dye formulation generates, is greatly enhanced work efficiency, save cost.

Computer for colouring technical development is provided to greatly the color matching of textile industry and develops, and the dyeing of existing computer is difficult to fundamentally solve the color matching problem of wood staining according to model. And due to the anisotropic of timber, cause obstacle also to the utilization of industry, it is clear that cannot effectively be applied in timber for conventional computer staining technique.

Summary of the invention

First the major anatomical factor tracheid that determining affects softwood wood staining effect is made a gesture of measuring, xylem ray is made a gesture of measuring, resin canal is made a gesture of measuring and the factors such as late wood's Tracheid Length; Affect the early wood conduit diameter of major anatomical factor timber of broadleaf wood staining effect, early wood fibre length, conduit are made a gesture of measuring, fiber proportion and xylem ray are made a gesture of measuring.

RBF neural is utilized to establish formula and technology model, and improve RBF neural model by a kind of method based on hidden node, the improvement of this model effectively solves the skilled slow problem of original model, and speed almost can reach the standard of on-line training. Finally fuzzy neural network is incorporated into forecast model, and proposes the membership function of a kind of improvement according to timber feature, establish based on the Fuzzy Neural Network Prediction Model improving membership function.

This model running speed is fast, and parameter arranges easily, it is not necessary to too much arranges condition, is greatly saved the time, and it is little to calculate error, and maximum error also only has 1.25%, illustrates that model has higher generalized ability.

The technology of the present invention feature:

(1) present invention uses the priori that technique is groped, and utilizes the intelligent algorithm such as neuroid, fuzzy algorithmic approach, sets up the imitative precious material dyeing intelligence color matching model of timber, stresses to utilize data fusion means that original algorithm is improved. And utilize multimodal data integration technology to be merged by above-mentioned model, in conjunction with multi-model information, set up the proprietary color matching model of final timber, to make wood staining formula output under the guidance of model more rapid, accuracy rate is higher, intelligent more powerful.

(2) from wood feature angle and dyeing recipe specificity analysis wood staining effect Changing Pattern, utilize Changing Pattern model and the combination of original model, make color matching system have intelligent predicting function.

The research of the present invention all has important practical significance for the raising of wood staining efficiency, the dyeing quickening of process of industrialization, the shortening of Recombinant Wood test period.

Detailed description of the invention

(1) wood staining intelligence color matching model is set up from wood feature angle.

First determine that the major anatomical factor tracheid affecting softwood wood staining effect is made a gesture of measuring, xylem ray is made a gesture of measuring, resin canal is made a gesture of measuring and the factors such as late wood's Tracheid Length; Affect the early wood conduit diameter of major anatomical factor timber of broadleaf wood staining effect, early wood fibre length, conduit are made a gesture of measuring, fiber proportion and xylem ray are made a gesture of measuring, and utilize fuzzy neuron network to set up intelligence color matching model afterwards.

(2) RBF neural is improved

Use the RBF neural improved based on hidden layer, the determination aspect of parameter is improved.

(3) membership function of fuzzy neural network is improved

Feature according to timber, uses and improves membership function, sets up based on the Fuzzy Neural Network Prediction Model improving membership function.

Claims

1. the computer for colouring method of the imitative treasure's dyeing of a novel timber, it is first determined affect softwood timber and the major anatomical factor of broadleaf wood staining effect, and it can be used as computer for colouring reference to affect data. RBF neural is utilized to establish formula and technology model, and improve RBF neural model by a kind of method based on hidden node, finally fuzzy neural network is incorporated into forecast model, and the membership function of a kind of improvement is proposed according to timber feature, establish based on the Fuzzy Neural Network Prediction Model improving membership function.

2. the computer for colouring method of the imitative treasure's dyeing of a kind of novel timber according to claim 1, it is characterised in that the design of computer for colouring technology and improvement:

(1) intelligent algorithm is introduced the prediction of wood staining color formulas.

(2) wood staining intelligence color matching model is set up from wood anatomy architectural feature angle.

(3) adopt a kind of mode based on hidden layer to improve RBF neural, and with regard to the determination aspect of parameter, improved.

(4) improve the membership function of fuzzy neural network, establish based on the Fuzzy Neural Network Prediction Model improving membership function.