Research On Reproduction-oriented Multispectral Image Acquisition Technology

Multispectral image acquisition technology is research focus in photographic, remote, digital archives and color reproduction, the multispectral image acquisition oriented to reproduction requires the image date is device independent and scenes independent, and it can realize the characterization of the original color information. This paper focus on several key techniques of high precision multispectral actuation in complex scene, including the environmental framework, surface illumination nonuniformity correction, system noise correction, spectral reconstruction, the training samples selection and so on, The experiments prove that the result is good.Investigated the technical advantages and system structure of the art reproduction based on multispectral, and summarize the technology development status and trend, then analysis the concepts and limitations of traditional image reproduction technology, pointed out that the "metamerism" is the biggest bottlenecks in current technical development. Last, discuss the technical advantages and the key technical problems in high fidelity reproduction based on multispectral.Investigated the device independent, scene independent image acquisition method, analysis the complex scenes and imaging system response characteristics effected the spectral information characterization. Discuss the effects of the light source properties on image acquisition, especially on the light source stability, color rendering, and we get the light sources spectrum characteristics requirement of multispectral image acquisition. Summarize the impacts of surface illumination error on the data precision, and proposed the illumination error correction algorithm Based on Chebyshev polynomial. Established the corresponding correlation function between CCD response value and exposure time based on the analysis of the CCD response characteristics, and investigated the spectral response characteristics. The imaging system spectral response was measurement by experiment; also it was calculated by mathematical model.Study on the imaging equipment noise error correction, Summarized the imaging system noise types and diffusion mechanism, aiming at the characteristic of dark current noise, Proposed the effective dark current correction method based on the pixel correction algorithm.Study on the spectral reconstruction algorithm, summarized two types of traditional method, according to the limitation of spectral reconstruction accuracy and Chroma error, and proposed the adaptive iterative spectral reconstruction algorithm, through the samples clustering and iterative operation, improving spectral reconstruction precision; according to the imaging system noise, proposed the orthogonal regression spectral algorithm, the experiment proves that this method can improve the spectral reconstruction accuracy.Study on the training samples selection algorithm, reveals the relation between the broadly representative, correlation and the reconstruction accuracy, proposed a new training samples selection algorithm based on subspace tracking, through the overall sample clustering, established the broad representation training samples selection mechanism, through the mapping function between the spectral space and color space, proposed the subspace tracking based training samples selection algorithm, by selecting the chrominance correlation maximum training samples, the reconstructed reflectance spectral and color precision were greatly improved...