基於影像復原之柵欄移除技術
No Thumbnail Available
Date
2007
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
影像復原，是指將影像中被遮蔽物所遮蔽的區域，加以填補的動作。其主要目的是希望能復原出原先被遮蔽的影像資訊。此技術可應用在老舊照片折痕的修補、照片上文字的去除、欄杆的移除等。本研究即是提出了一個基於影像復原之柵欄移除技術，自動化填補未知區域的影像資訊，從一張靜態的數位影像上，移除前方的柵欄，顯現出柵欄後面的影像資訊。
柵欄移除技術，主要是利用兩大步驟：邊緣偵測和漸進演算法。
第一步驟，利用索貝爾(Sobel)的邊緣偵測特性，將影像中邊緣線找出來；接著利用柵欄的性質，對柵欄進行估測，刪除不是柵欄的邊緣及雜訊，以及修補斷掉的柵欄邊緣線，呈現出影像中柵欄的位置。
第二步驟，對已判別出柵欄的區域進行移除，並修補原先被柵欄遮蔽的影像資訊，此方法藉著柵欄區周圍的資訊，進行顏色漸進的效果來達到復原影像。
利用上面的兩大步驟，可以對單張靜態的數位影像，進行柵欄移除，並對原先被遮蔽的部分進行復原影像資訊，而呈現出無柵欄的完整影像。
Image inpainting is to fill in the areas hiding from view. Its main purpose is to recover out the image information originally covered. The technology can be applied to old photos that have some creases, and get rid of characters, move fences, etc. on photos. Research here is a method based on the image inpainting technology that can remove the fence on a static photo. It can automatically fill in image information of unknown area. It can remove the fence in front of a piece of static image. Then it displays image information behind the fence. The technology of removing fence mainly utilize two major steps: detect the edge and a step by step color algorithm. The first step, it utilizes the characteristics on Sobel Edge Detection to find out the edge line in an image. Aftter that, it utilizes the nature of the fence to delete the edge and miscellaneous noise that is not belonged to the fence. It also repairs the fence edge line shed to break. Then it demonstrates the position of the fence of the image. Second step, it removes the area of fences that have already found. It recovers image information that covers by fence. The method use the step by step color algorithm to reach the image of inpainting by information around the fences. Utilizing two major steps above, it can move the fence from a single and static image and recover the information under the fence.
Image inpainting is to fill in the areas hiding from view. Its main purpose is to recover out the image information originally covered. The technology can be applied to old photos that have some creases, and get rid of characters, move fences, etc. on photos. Research here is a method based on the image inpainting technology that can remove the fence on a static photo. It can automatically fill in image information of unknown area. It can remove the fence in front of a piece of static image. Then it displays image information behind the fence. The technology of removing fence mainly utilize two major steps: detect the edge and a step by step color algorithm. The first step, it utilizes the characteristics on Sobel Edge Detection to find out the edge line in an image. Aftter that, it utilizes the nature of the fence to delete the edge and miscellaneous noise that is not belonged to the fence. It also repairs the fence edge line shed to break. Then it demonstrates the position of the fence of the image. Second step, it removes the area of fences that have already found. It recovers image information that covers by fence. The method use the step by step color algorithm to reach the image of inpainting by information around the fences. Utilizing two major steps above, it can move the fence from a single and static image and recover the information under the fence.
Description
Keywords
柵欄移除, fence removing