Author | Conference | Journal | Organization | Year | DOI
Look for results that meet for the following criteria:
Academic
Publications
Region filling and object removal by exemplar-based image inpainting

Region filling and object removal by exemplar-based image inpainting,10.1109/TIP.2004.833105,IEEE Transactions on Image Processing,Antonio Criminisi,P

Edit
Region filling and object removal by exemplar-based image inpainting   (Citations: 342)
BibTex | RIS | RefWorks Download
Abstract, A new algorithm is proposed for removing large ob-jects from digital images. The challenge is to fill in the hole that is left behind in a visually plausible way. In the past, this problem has been addressed by two classes of algorithms: 1) "texture synthesis" algorithms for generating large image regions from sample tex-tures and 2) "inpainting" techniques for filling in small image gaps. The former has been demonstrated for "textures", repeating two-dimensional patterns with some stochasticity; the latter focus on linear "structures" which can be thought of as one-dimensional patterns, such as lines and object contours. This paper presents a novel and efficient algorithm that combines the advantages of these two approaches. We first note that exemplar-based texture synthesis contains the essential process required to replicate both texture and structure; the success of structure propagation, how-ever, is highly dependent on the order in which the filling proceeds. We propose a best-first algorithm in which the confidence in the synthesized pixel values is propagated in a manner similar to the propagation of information in inpainting. The actual color values are computed using exemplar-based synthesis. In this paper, the simultaneous propagation of texture and structure information is achieved by a single, efficient algorithm. Computational efficiency is achieved by a block-based sampling process. A number of exam-ples on real and synthetic images demonstrate the effectiveness of our algorithm in removing large occluding objects, as well as thin scratches. Robustness with respect to the shape of the manually selected target region is also demonstrated. Our results compare favorably to those obtained by existing techniques. Index Terms, Image inpainting, object removal, simultaneous texture and structure propagation, texture synthesis.
Journal: IEEE Transactions on Image Processing , vol. 13, no. 9, pp. 1200-1212, 2004
Cumulative Annual
View Publication
The following links allow you to view full publications. These links are maintained by other sources not affiliated with Microsoft Academic Search.
More »
    • ...Criminisi et al. proposed an inpainting method for region filling and object removal [7]...

    Chuan Qin, et al. Simultaneous inpainting for image structure and texture using anisotro...

    • ...A number of user-driven [12] and intelligent [3] heuristics have been proposed for achieving structural image coherence...

    Alex Mansfield, et al. Transforming Image Completion

    • ...This strategy (with various modifications) has been extensively used for inpainting (Efros and Leung 1999; Bornard et al. 2002; Bertalmío et al. 2003; Drori et al. 2003; Criminisi et al. 2004; Pérez et al. 2004)...
    • ...Fig. 1). The unknown portion of the image is then synthesized using the correspondences Γ . The filling-in strategy of Efros and Leung (1999), Wei and Levoy (2000) can be regarded as a greedy procedure (each hole pixel is visited only once) for computing a correspondence map .T he results obtained are very sensitive to the order in which the pixels are processed (Criminisi et al. 2004; Pérez et al. 2004; Harrison 2005)...
    • ...For large inpainting domains, it is useful to introduce a mask κ : Ω → (0, 1] which assigns a confidence value to each pixel, depending on the certainty of its information (see also Criminisi et al. 2004; Komodakis and Tziritas 2007)...

    Pablo Arias, et al. A Variational Framework for Exemplar-Based Image Inpainting

    • ...The most popular completion methods use patches from other locations in the image or video as source information for synthesizing data inside target holes (e.g., [8], [9], [23], [38], [42], and [43])...
    • ...In most of these methods (e.g., [8] and [9]), the completion process is sequential, propagating the fill in from the boundary of the known data domain, into uncharted territories...
    • ...Geisler and Perry [12] used a Gaussian pyramid [6] of the input image . Each pyramid level corresponds to Fig. 5. Core completion engine [8]...
    • ...As illustrated in Fig. 3, our foveated video extrapolation algorithm (which we describe in Section IV) uses a full resolution video completion algorithm as a black-box “engine.” We suggest a specific engine (Section III) which follows the patchbased algorithm suggested by Criminisi et al. [8]...
    • ...Criminisiet al.[8]. Our description is in the context of video extrapolation...
    • ...The ST blocks1 are of constant size of . The target block is always on the boundary between and . Blocks that include strong ST edges crossing , and those that include less missing pixels, 1 Following [8], the blocks are boxes...
    • ...Most of the computation time in each iteration is spent on searching for a source block . The original algorithm [8], [24] searches the entire . Some methods reduce the search domain, either by filtering out irrelevant candidates using simple statistics (e.g., [9]) or by a prior semantic analysis of the scene (e.g., [20]‐[22] and [28])...

    Tamar Avraham, et al. Ultrawide Foveated Video Extrapolation

    • ...Moreover, many contributed works have been proposed for the solution of this interpolation task based upon (a) diffusion and transport PDE/variational principle [6], [13], [23]‐[25], [34], [64], [70], (b) exemplar region fill-in [12], [28], [30], [31], [65], [76], (c) compressive sensing [32]...

    Miyoun Jung, et al. Nonlocal Mumford-Shah Regularizers for Color Image Restoration

Sort by:

Citations (342)

Share this on