近年來,數位影像處理的技術日趨成熟,有許多的數位產品都附有照像功能。例如數位相機、行動電話、筆記型電腦和PDA 等。數位影像裝置儼然已成了日常生活中不可缺少的必須品,也改變了我們的生活。其中數位相機更被廣泛地用來當作影像的擷取裝置,也漸漸取代了傳統底片的相機。 在我們的生活中由於數位彩色影像的普及,因此消費者對於影像品質的要求也愈來愈大。在影像中,色彩的呈現對於影像品質是重要的條件之一。正確的色彩能提供較好的品質,也能讓影像更趨真實。然而,當我們要拍攝某件物體時,物體往往會受到任何光源的影響,例如,太陽或者是街上的路燈。為了要改善光源所帶來的影響,我們提出一個簡單又有效的方法。 在這篇論文中,我們利用一個灰階調修正的色彩校正方法是藉由灰界值理論學說所衍生出來的。而所提出來的灰階調修正法可以有效地將一張被不知名光源所影響的影像,校正到標準色溫(D65)底下。而影像經過了色彩校正之後,再利用色域對應的方法來提升彩色的精確度。而色域對應矩陣也代表著閃燈模式與自動白平衡之間的色彩關係。而經由不同的色差臨界值可以得到不同效果的處理影像。最後我們再從這些數十張處理後的影像經由色相與色度的分析來尋找到一個最適當的處理影像。而這個最適當的影像也確實符合我們所利用的灰界值理論。經由實驗結果可以顯示出我們所提出來的色彩校正方法在不同的地點與不同的內射光源下,就算影像裡有大範圍相同或相似的色彩 都能有很好的校正效果。
Technology of digital image processing is becoming more and more mature recently. There are a lot of digital products append photograph function, such as digital camera, mobile phone, notebook, PDA, etc. Digital image devices have become essential in people’s life and changed the way of our living. Digital cameras have been widely used as image input devices and replaced the traditional film cameras gradually. Due to the digital color image become quite popular in our life, hence the requirements of achieving consumers’ expectations about the image quality have broadened gradually. In an image, the color appearance is one of the important conditions. Correct color could prove a better quality and make the images to get a more accuracy. However, when we are taking the image, the objects are always affected by any illuminants such as the sun or some lamps in the streets. In order to modify the influence of illuminant, we proposed a simple and effective approach. In this thesis, we utilize the color calibration method that derived from the Gray World Assumption - Gray Tone Correction. The proposed Gray Tone Correction could adjust the image that effected by an unknown incident light source into the standard color temperature (D65) effectively. After the step of color correction, next is to use a Gamut Mapping Matrix method to improve the color accuracy. The Matrix of the Gamut Mapping represents the color relating to the Flash mode and auto white balance mode. Due to the different color deviation threshold, we could obtain the different result from the processed image. Thus, we search the fittest image in these several tens processed image by hue and chromaticity analysis. The fittest image that we found actually conformed to the Gray World Assumption. By the experimental results, it shows the proposed color calibration actually performs well in any different place and different incident light sources even if there is a lot of region with same or similar colors.