Theory And Practice Of Color Fidelity Replication

In recent years, the traditional color reproduction method can not meet the needs ofhigh-end works of art reproduction, high-end luxury goods packaging on color separationprecision and color gamut expansion, which promote the progress of high fidelity colorreproduction and lead to technology innovation on the field of printing industry. Improving theprecision of spectral-based separation and extending the gamut of traditional four-colorreproduction are key for high fidelity color reproduction. By investigating spectral separationmethod based on neutral network and basis color prediction method based on characteristicspectra extraction, the study aims to construct a high fidelity reproduction model ofmulti-primary color and high separation precision, so as to effectively promote the industrialapplication of high fidelity reproduction. The results are not only important for the field of highfidelity reproduction, but also have vast importance in spectral discrimination and matching,multi-spectral image compression, and medical image analysis and diagnosis.The main research results are as follows:1. Constructs spectral separation models based on BP and RBF neutral networkrespectively; reduces color separation error by spectral density transformation; optimizes thestructure and parameters of neutral network. The experiment proved that the spectral separationmodel based on neutral network is efficient and of high separation precision.2. Introduces an extraction algorithm for characteristic spectra based on derivativetransformation and correlation analysis, and the experiment proved that the characteristicspectra of traditional C, M and Y ink are mainly concentrated on the430nm,440nm,490nm,510nm,550nm,560nm,590nm,650nm and700nm bands, under the condition of10nmprecision of measurement.3. Introduces an algorithm for basis color prediction based on characteristic color andcharacteristic spectra, experiment proved that characteristic color and characteristic spectraextraction can effectively reduce the number of samples, and improve the precision of basiscolor prediction. It is also proved by gamut simulation that the predicted basis colors can extendthe traditional4-color gamut.4. Introduces an algorithm for spectral separation based on neutral network andcharacteristic spectra matching. The experiment proved that basis color extraction and spectralseparation based on characteristic spectra matching can effectively improve the precision ofchromaticity matching without losing spectral matching precision.5. Based on above researches, the study constructs a high fidelity color reproduction modelof multi-primary color with high separation precision, which focuses on neutral network andcharacteristic spectra extraction and matching. The simulation showed that the model canmaximize the extension of gamut, and had high separation precision.