自然影像中的光譜估計
No Thumbnail Available
Date
2009
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
彩色影像的呈現是由物體反射譜或透射譜、光源的光譜分佈及人眼感知三項變數交互作用而成,在拍攝過程中,會因為環境光源的變化導致影像中物體的顏色失真,想如實地呈現影像的色彩就必須掌握拍攝時環境光源的狀況。本研究的目標是希望可以利用色彩學的概念和統計學的方法來估計影像拍攝時環境光源的頻譜,以利模擬出接近物體原始色彩的影像。
基於上述的動機,本研究計畫利用主成分分析(Principal Component Analysis)、支援向量回歸(Support Vector Regression)兩種方法來重建影像中日光的光譜,並比較兩個方法在光譜重建上的效果。希望能藉由實驗結果來提高影像顯示的準確性,並應用於影像合成、數位典藏等領域,讓我們可以透過修正影像中的光源頻譜,來獲取更佳的影像表現。
The color image is the result of interaction among the reflectance of object, the spectra of illuminant incident on the scene and the eye. The change of environmental light source leads to the object show a different color during the process of capture. If we want to present true color, we must control environmental light source. The goals of the study focus on using the conception of color technology and method of statistics to estimate the illuminant spectra of environment, and rebuilding the original color image. In this study, we use Principal Component Analysis(PCA) and Support Vector Regression (SVR) to rebuild the spectra of daylight. finally we will assess the effect of two approach in spectra rebuilding. We hope through experimental results to improve the accuracy of image display. These results will be applied to color adjustment of synthesis image, digital archives, medical imaging and other digital image post-production process to make it be more natural.
The color image is the result of interaction among the reflectance of object, the spectra of illuminant incident on the scene and the eye. The change of environmental light source leads to the object show a different color during the process of capture. If we want to present true color, we must control environmental light source. The goals of the study focus on using the conception of color technology and method of statistics to estimate the illuminant spectra of environment, and rebuilding the original color image. In this study, we use Principal Component Analysis(PCA) and Support Vector Regression (SVR) to rebuild the spectra of daylight. finally we will assess the effect of two approach in spectra rebuilding. We hope through experimental results to improve the accuracy of image display. These results will be applied to color adjustment of synthesis image, digital archives, medical imaging and other digital image post-production process to make it be more natural.
Description
Keywords
光源頻譜, 光譜估計, 主成份分析, 支援向量回歸, Illuminant spectra, Illuminant estimation, Principal Component Analysis, Support vector regression