Research On Composition Analysis Method Of Ancient Mural Mixed Pigments Based On Spectral Reflectance

Recognizing the mural paint is an important research content in the field of cultural relics protection,and is also an important prerequisite for cultural relics protection workers to repair murals.In the ancient murals,mixed pigments were used.Component analysis of mixed pigments is an important research topic in the field of murals.Combining rapid independent component analysis and reflection spectroscopy is an effective model for unmixing mixed pigment spectral information.Based on the characteristics of the pigment reflection spectrum curve,this dissertation studies the unmixing model of different pigment mixing conditions.The main contents include:(1)For the "S"-shaped pigment mixing model,an unmixing algorithm based on cluster optimization of fast independent component analysis for mixed pigment spectral information is proposed.In the process of spectral information unmixing,the fast independent component analysis is sensitive to the initial value of the unmixing matrix,which leads to the instability of the unmixing information and the low accuracy of the unmixing.A fast independent component analysis unmixing algorithm based on cluster optimization is studied.The fuzzy Cmeans clustering algorithm is used to reduce the spectral characteristics of the single-color pigment spectral information,and the most representative clustering result is selected as the initial value of the unmixing matrix,which is beneficial to the enhancement of the unmixing performance and the improvement of the unmixing accuracy.The experiment selects the pigment whose reflection spectrum curve conforms to the "S" shape mixing model to make the mixed pigment samples,and unmixes and analyzes a group of real mural pigments.The experimental results show that the optimized unmixing algorithm increases the stability of the mixing result and improves the unmixing accuracy of the spectral information of the "S" mixed pigment.(2)For the "bell"-shaped pigment mixing model,an unmixing algorithm based on Gabor transform of fast independent component analysisfor mixed pigment spectral information is proposed.For the problem that independent component analysis can not directly unmix multiple Gaussian distribution source signals,an improved fast independent component analysis unmixing algorithm based on Gabor transform is studied.Use the Butterworth filter to remove the common frequency in the spectral data of the mixed pigment after Gabor transform to eliminate the influence of the spectral information of the monochromatic Gaussian pigment on the unmixing results;And then combine Cluster optimization and fast independent component analysis unmixing algorithm to unmix the transformed mixed pigment spectral information.Experiments on the "bell" mixed pigment samples verify the feasibility of the algorithm.The experimental results show that the algorithm has less error in the unmixing results and effectively improves the accuracy of the unmixing.