Fabric image color calibration algorithm based on spectral reflectivity reconstruction

Abstract

The invention discloses a fabric image color calibration algorithm based on spectral reflectance reconstruction, relates to the field of fabric image color correction, and solves the problem that thereconstruction precision is affected by image acquisition color distortion in the existing method. The technical scheme includes: acquiring color training samples through collection; constructing a virtual imaging system, and calculating virtual system responses of all color training samples and corresponding spectral reflectance estimated values; performing color space conversion according to a color space conversion formula; adopting a defined spectrum reconstruction error evaluation function to determine a representative color training sample; obtaining a spectral response function of an actual imaging system according to the determined representative color training sample; and obtaining a corresponding reconstructed spectral reflectance through a Wiener estimation method, and obtaininga reconstructed RGB image through color space conversion. According to the fabric image color calibration algorithm based on spectral reflectance reconstruction, the quality of the fabric image is improved, and real yarn color information of the fabric image is obtained.

Description

technical field [0001] The invention relates to fabric image color correction technology, in particular to a fabric image color calibration algorithm based on spectral reflectance reconstruction. Background technique [0002] In the process of fabric production and analysis, digital image analysis technology is used to extract features and design patterns of fabric images, all of which are based on the color data of fabric images. However, because machine vision lacks adaptability and intelligence, and does not have the characteristics of color constancy, changes in image acquisition, transmission, and display systems will cause changes in image color data, directly affecting the accuracy of analysis results. Therefore, it is particularly important to make necessary corrections to the color data of fabric images. [0003] The research shows that the correction method of spectral reflectance reconstruction can effectively reduce the influence of light source and environment ...

AI Technical Summary

Problems solved by technology

[0004] However, these methods select representative color training samples based on the minimum average spectral mean square error as the judgment condition. Although they reduce the redundancy of color samples and also show good spectral accuracy, the study found that when the spectral mean square error is the smallest , the color difference of the reconstructed color may not be the smallest, so there is no guarantee that the representative color training samples selected by these methods can have ideal reconstruction accuracy

Images