基于全局优化的图像块填充修复方法.PDFVIP

ISSN 1000- 1239PCN 11- 1777PTP Journal of Computer Research and De elopment 46( 1) : 144- 150, 2009 陈中贵 刘利刚 王国瑾 ( 310027) ( CA D CG 310027) ( ligangliu@ zju. edu. cn) Exemplar-Based Image Completion Using Global Optimization Chen Zhonggui, Liu Ligang, and Wang Guojin (Dep artment of Mathematics, Zhej iang Univ ersity , H angzhou 310027) (State K ey Laboratory of A D G, Zhej iang Univ ersity , H angzhou 310027) Abstract Image completion, w hich aims to remo e objects or reco er the damaged portions in a gi en image, is an important task in photo editing. Recently, exemplar-based methods are consider ed to complete images w ith large portions remo ed. How e er, structure inconsistency of the reconstructed texture often appear w hen using those methods. In this paper, a new exemplar-based algorithm is proposed to obtain global texture consistency by using global optimization. First, an energy function is defined for measuring the quality of the reconstructed region. Then, the image completion problem is formulated as minimization of the energy function w hich is done in an iterati e form. Finally, the slight color differences betw een the know n r egion and the filled region are re ised by the Poisson image editing method. Compared w ith the existing exemplar-based methods w hich do greedy region- grow ing , the proposed method not only reconstructs the local color texture of missing region, but also preser es the global structur al texture of the image. An adapti e sampling method, w hich is based on the saliency map of the image, is also adopted to construct the sear ching space. It dramatically reduces the searching space and accelerates the nearest neighbor sear ching . The effecti eness of the proposed method is demonstrated on se