以色外貌模式為基礎高動態範圍顯示器色彩修正技術之研究

Abstract

我們的人眼視覺系統（Human Visual System，HVS）基本上是一組幾近完美的高動態範圍（High Dynamic Range，HDR）影像擷取系統，其視明度（Brightness）之可處理範圍非常大，同時又具備非常優異的視覺適應（Adaptation）能力，但是對於一般常用的影像擷取與顯示設備，只能在極為有限的動態範圍間運作，導致有許多影像無法清楚被呈現，為了解決這樣的問題，開發HDR顯示技術在現今普遍重視高畫質影像顯示的時代變得非常重要。 高動態範圍合成是一種可以擴大亮度範圍的影像處理技術，但儘管目前有許多HDR的演算法被開發出來，但是這些方法在不同曝光程度影像合成時容易產生色調不協調或色調偏移的現象，特別是對於相同場景而不同環境光源照射下拍攝時色溫（Color Temperature）不一致情形特別嚴重，儘管影像資訊能夠清楚呈現出來，但是色調卻已經失真，這對高品質影像顯示而言是一項非常嚴重的缺失。 本計劃將利用色外貌模式（Color Appearance Model，CAM）對影像進行色彩調整，預計提出一個HDR影像合成與色彩校正的流程，主要的研究課題包括以下三個部分，一是以多張不同曝光程度的數位影像進行HDR影像合成，以解決不同環境光源、不同曝光程度的影像在明度、彩度、色相上偏移所造成合成時的失真問題。二是透過CAM對數位影像進行色彩調整，我們利用Fairchild與Johnson於2004年發表的iCAM（Image Color Appearance Model）進行光源轉換與色彩調整，使影像能顯示準確的色彩。三是針對顯示器特性的適應性調整，期望達到數位影像在不同顯示器上獲得最佳的色彩效果。預期本研究的成果能作為影像顯示設備設計時的重要參考，以促進高畫質顯示製造產業的發展。

The human visual system basically tends to a perfect image capture system with high dynamic range (HDR). The working range of brightness diverse much larger, and it also has a great visual adaptive ability. However, most image captured or display devices now can work with limited dynamic range only; therefore, cause the disappearance of many details of images. HDR image compositing is an image processing method that can extend the dynamic range of multiple images. Although there are many previous works studying their detail preservation by the tone mapping methods proposed for various exposure levels of images, they still lead to an another problem of color consistency. The brightness, saturation, or even hue has truly been distorted seriously especially for those images captured under various illuminants. That is a serious drawback for the recent display requirement of high quality. In this project, the color appearance mode (CAM) will be used to adjust the color of images. We propose a working flow of color correcting process during the HDR compositing of multiple images. This research includes three major topics: Firstly, we will design a new HDR compositing method to solve the color distortion problem caused by the images captured at different exposure level and under different lighting environments. Secondly, the iCAM proposed by Fairchild and Johnson will be modified to develop the method of illuminant simulation and color correction in the compositing image. The third, the adaptation according to the display characteristics will be also proposed to obtain the optimal color appearance effect on the displays of various kinds. It is expected that the results obtained in this project can serve as an important reference to improve the design of image displays of high quality.