利用lαβ色彩空間達成灰階影像色彩化

Abstract

本篇論文是利用lαβ這種貼近人類視覺習慣且三個色彩頻道極少關聯性的色彩空間來研究灰階影像的上色與影像色彩的轉移。儘管灰階影像的上色並沒有「唯一」的正確答案，本論文希望提出一套方法，能盡量減少需要人為手動參與的部分。不同於以往需要人們為每一個灰階自行定義所對應的色彩，本論文以影像分割為基礎提出一個有效率的色彩轉移方法，藉由比對亮度以及紋理特徵，將所希望的色調完全轉移到灰階或彩色影像。藉著快速影像分割以及有效的比對技術，使得灰階影像的上色速度獲得大幅的提升，也讓其更具備實用性。

We take advantage of the uncorrelated lαβcolor space to accomplish colorization of grayscale images. Even though the problem of adding chromatic values to a grayscale image has no exact solution, our approach attempts to provide a method to minimize the amount of human labor required for this task. Instead of choosing RGB colors from a palette to color individual components, we propose a segmentation-based color transferring method to deliver entire color mood to target images by matching luminance and texture information between the images. With a novel time-saving image segmentation procedure and an effective color-matching method for grayscale images, we get substantially speed-up and make it possible to be used in video processing applications.