Research On Detection Method Of Wood Dyeing Quality Based On Spectral Similarity

Dyeing quality inspection is an important part of the process of dyeing wood veneer to imitate precious wood.In the color matching process,accurate measurement of the color difference between the color of the dyed wood and the target wood is the basis for establishing a dyeing formula model.In the process of quality inspection,accurate measurement of the wood color and color fastness of dyed wood is an effective basis for product qualification inspection and quality grading.In the traditional wood dyeing industry,spectrophotometry is mainly used for color measurement.However,the spectrophotometer only can perform offline single-point measurement,the measurement efficiency is low,and the one-dimensional point measurement cannot feedback the specific position of the measurement wood plane where the measurement point is located,thereby restricting the development of the wood dyeing industry.From the perspective of spectroscopy and chromaticity,this paper uses the spectral characteristics of the surface of the object to realize color difference detection,and achieves"online integration" of pixel-level dyeing quality on-line detection.This paper proposes a color difference detection method based on spectral similarity.This method uses a hyperspectral imaging system to obtain the spectral reflectance of the object surface.Study the relationship between the characteristics of the spectral reflectance curve and the color of the object,analyze the influence of characteristic parameters of spectral reflectance curve on lightness,chroma and hue.Euclidean distance measurement,spectral angle measurement,correlation coefficient measurement,spectral information measurement and spectral angle cosine-Euclidean distance measurement were used to measure the spectral similarity between standard color card samples,and the spectra were analyzed.Consistency of similarity and color difference in color difference evaluation.Taking the difference in spectral characteristic parameters between samples as input,the color information chromaticity parameters ΔL,Δa*,Δb*between samples are output based on the construction of a color difference detection model based on spectral similarity.The genetic algorithm was used to optimize the connection weights and node thresholds of the extreme learning machine model,and compared the ΔL,Δa*,and Δb*value prediction accuracy and color difference detection error of the model when the five spectral similarity measurement methods described above were used as target errors.The experimental results show that the spectral angle cosine-Euclidean distance is established as the error target,and the color difference detection model based on spectral similarity has the best detection accuracy.In this paper,reactive red X-3B dye,reactive brilliant blue dye,and reactive yellow X-RG dye were used to prepare wood stained veneers as experimental samples.In this paper,reactive red X-3B dye,reactive brilliant blue dye,and reactive yellow X-RG dye were used to prepare dyeing wood veneers as experimental samples.Deteriorate it by self-made UV radiation deterioration experiment box.The color difference detection and color fastness analysis were performed on the wood that had been deteriorated by ultraviolet radiation for 100 hours.The experimental results show that the color difference detection model based on spectral similarity established in this paper can accurately predict the color difference of dyed wood.Dyeing quality detection method based on spectral similarity provides new research ideas for color difference detection and color fastness evaluation of dyed wood.And based on the wood staining quality detection method and model proposed in this paper,the design of wood staining quality detection system is realized by MATLAB programming,which provides convenience for the industrial application of research results.