Channel Selection And Software Design For A Multispectral Imaging System

The multispectral imaging system combines the advantages of both camera and spectrophotometer, which is a spectral image measurement instrument capable of measuring spectral reflectance with a high spatial resolution. Additionally, it is carefully designed to ensure the very spectral accuracy and repeatability.To effectively overcome the problem of speeding up the image acquisition process and keeping reflectance reconstruction accuracy, it is of significant importance to select an optimal subset of imaging channels. In this paper, the optimal channel selection problem is converted to a binary optimization problem, which is, consequently, solved using a novel binary differential evolution (DE) algorithm. In the proposed algorithm, tables of swapping pairs and sefl-crossover are used for the reason of the fact that the binary DE algorithm can be suitable to the channel selection problem. The performance of the proposed algorithm, evaluated on the multispectral imaging system with both synthetic and real data, is better than that of the traditional sequential channel selection.This paper, as well, provides one-stop solution of software design for a multispectral color imaging system including framework analysis, user interface design, operation procedure and other related design. Furthermore, it divides the software system into multispectral imaging module, spectral data processing module and additional functions module, whose design and realization are elaborated in details. The designed system, as shown in the paper, evolves the field of textiles to evaluate and process the spectral reflectance data of fabric samples ranging from yarn-dyed samples, laces to printing fabrics effectively.