A Intelligent Computer Color Matching Technology In Wood Dyeing

Aim at the surface painting production process for Huzhou wood flooring business,this paper were study on one of the most core technical difficulties in wood dying-color matching.Use of of colorimetry, integrated optics, visual psychology, image processing and information processing technology, a combination of traditional color theory and information technology, and developed a fast, accurate and intelligent wooden floor computer color measurement system, research hascertain practical significance.Conform to the the intelligent modern color system requirements,and combined with actual production, this paper compared the effect of the neural network and support vector regression, selected the support vector regression machine which has better generalization ability as the intelligent technology of the color system.At the same time according to the actual production needs, doing works of data analysis and feature extraction, established a series of database data. Based on user demand,build up a set of easy-to-use computer color matching system for wood flooring including both hardware and software.The paper compares several image acquisition in accordance with the implementation process.Introduces several common color space, and the histogram color feature extraction method.Selects two excellent regression model:generalized regression neural network (GRNN)and support vector regression machine(SVR)made comparative study.Discusses the multi-output support vector regression algorithm in view of the traditional support vector machine’s one-dimensional output.Analyze the data come form the data collection and the establishment of multiple databases.According to the difference between different color scheme, puts forward a new color scheme difference algorithm.The test results show that this algorithm is more visual uniformity than the traditional Euclidean distance.The algorithm was introduced into the model parameter optimization objective function,comparison of two kinds of parameter optimization algorithm:grid search and particle swarm algorithm,trained8regression models,that each were tested to compare their results.In the end,the algorithm based on RGB color space feature vector multi-output support vector regression were selected for color matching.By Visual C#and matlab mixed programming technology, a wood floor computer color measurement-matching system were completed.The experimental results and the performance analysis shows that, the matching system can successfully deal with the small sample, nonlinear problem in the process of wood floor dyeing, the generated color scheme data come form the matching system were proved to be reliable, and good.The system has enhanced wood floor production process automation, reduces the artificial color time, reduces the color of the instability of the process, improve efficiency..