應用於手持裝置之影像物件置換演算法

Image Object Replacement Algorithm Applied to Mobile Devices

王偉諠(Wei-Syuan Wang)

指導教授 ： 劉震昌

國立暨南國際大學/科技學院/資訊工程學系/碩士(2011年)

摘要

智慧型手機與平板裝置在最近迅速發展，許多手持裝置的應用軟體也大量出現。表示行動運算以融入日常生活中。大部分智慧型手機都有相機的功能，使得手持裝置上的影像處理相關應用逐漸受到重視，包括影像編輯、影像修補，影像特效等。發展手持裝置上的影像處理演算法的困難之處在於要考慮手持裝置上有限的資源與處理器速度。
本論文提出建構於行動裝置上的影像物件置換演算法與創新應用。以人物為置換目標，可以達到趣味角色扮演的功能，也可以針對任何物件來進行置換。透過行動裝置上的照相機與觸控螢幕，系統提供使用者第一次拍照並圈選欲置換的物件區域，接著將欲置換的區域挖空並顯示在螢幕上，讓使用者可以對準要換入的目標物件並二次拍照，應用程式會將目標物件與第一張影像的背景合成，可提供使用者互動式的趣味影像編修應用。此演算法使用輪廓提取技術，將使用者圈選的輪廓自動校正，符合真正物件的輪廓；利用影像修補，自動修補目標物件與背景之間的空隙；利用顏色轉換，讓目標物件的影像色調更相近於欲置換的物件。
本論文的實驗分別為輪廓提取、影像修補與顏色轉換。我們使用大量的範例來驗證提出的影像物件置換演算法的效果。另外，影像修補是實作在手持裝置上，所以在實驗方面有針對大範圍與小範圍的修補區域，實際在手持裝置上測試影像修補的執行時間。

關鍵字

物件置換 ； 輪廓提取 ； 影像修補 ； 顏色轉換

並列摘要

The applications of mobile devices have been rapidly developed on smart phones and tablets in recent years. Along with the rapid growth of the Internet community over the past decade, Mobile Computing has become a part of our everyday lives. Most mobile devices have built-in cameras. Therefore, the image processing applications on mobile devices are receiving more attentions, such as photo editing, image inpainting, and special effects on images. The difficulties of developing image processing algorithms on mobile devices are due to the limited resources and processing speed on these embedded systems.
In this thesis, we propose an image object replacement algorithm on mobile devices. First, the user can take a photograph. On the touch screen, he/she can draw the contour around the object for replacement. The system will display the photograph on the camera view, where the region inside the contour will be transparent. Secondly, the user can take a photograph again by aligning the transparent region with the target object in the real world. After taking the second image, he/she can draw the contour around the target object. The application will replace the target object of the second image on the background of the first image to compose a synthesized image. The digital image processing technologies applied include contour extraction, image inpainting, and color transfer. Contour extraction is used to modify the drawn object contour by the user. Image inpainting is used to fix the gaps between the target object and the background. We use color transfer to transfer the color tone of the target object from the replacement object.
The experimental results on the contour extraction, image inpainting, and color transfer are presented. Enormous examples are examined to show the effects of the proposed image object replacement algorithm. Our image inpainting algorithm is implemented on mobile devices, and the execute time on the mobile devices is studied.

並列關鍵字

object replacement ； contour extraction ； image inpainting ； color transfer

參考文獻

[1] A. Criminisi, P. Perez, and K.Toyama, “Region Filling and Object Removal by Exemplar-Based Image Inpainting,” IEEE Trans. Image Processing, Vol. 13, No. 9, pp.1200-1212, 2004.

[2] A. Telea, “An Image Inpainting Technique Based on the Fast Marching Method,” Journal of Graphics Tools, Vol. 9, No. 1:25-36, 2004.

[3] A. Levin, D. Lischinski, and Y. Weiss, “Colorization using optimization,” ACM Trans. Graph., 23(3):689-694, 2004.

[4] A. Maslennikova, V. Vezhnevets, “Interactive Local Color Transfer Between Images,” GraphiCon, 2007.

[5] E. Reinhard, M. Ashikhmin, B. Goochm P. Shirley, “Color Transfer Between Images,” Computer Graphics and Applactions, Vol. 21, Issue 5, pp. 34-41, 2001.