ISSN (Print) : 2319-5940 ISSN (Online) : 2278-1021 International Journal of Advanced Research in Computer and Communication Engineering Vol. 2, Issue 1, January 2013 Copyright to IJARCCE www.ijarcce.com 964 Digital Image Inpainting Using Cellular Neural Network and Contour Tracking Using Run Length Coding Usha Kiran 1 , Om Prakash Yadav 2 Assistant Professor, Dept.of CSE, CSIT, Durg, India 1 Associate Professor 2 , Dept.of CSE, CSIT, Durg, India ,2 ABSTRACT: Digital Image Inpainting is challenging and interesting research area, because one has to restore the area which is not visible but important to visually complete the image. This technique has found widespread use in applications such as restoration, error recovery, multimedia editing, and video privacy protection. Because of the strong human visual perception, a very effective technique is required for digital image inpainting. Most automatic techniques are computationally intensive and unable to repair large holes. Existing methods use interpolation methods where surrounding information is not adequate for image interpolation and chain codes for contour matching for small damaged area reconstruction. But, reconstructed image has not given up to the mark results. This paper proposed an effective inpainting technique in order to improve the inpainting result. The method proposed in this paper uses Run Length Coding for shape tracking along with CNN approach, because Run Length coding track the shape of an image, which is better than the several methods available for shape tracking. Keywords: Image inpainting, Cellular Neural Network, Digital images, contour matching, Run Length Code I. INTRODUCTION Reconstruction of missing or damaged portions of images is an ancient practice used extensively in artwork restoration. This activity, also known as inpainting or retouching, consists of filling in the missing areas or modifying the damaged ones in a manner non-detectable by an observer not familiar with the original images. The goal of inpainting algorithms varies, depending on the application, from making the inpainted parts look consistent with the rest of the image, to making them as close as possible to the original image, restoration of photographs, films and paintings, to removal of occlusions, such as text, subtitles, stamps and advertisements from images. In addition, inpainting can also be used to produce special effects. While, traditionally skilled artists have performed image inpainting manually, currently digital techniques are used, e.g. for automatic restoration of scratched films. 1.1 DIGITAL IMAGE INPAINTING TECHNIQUES As a first step the user manually selects the portions of the image that will be restored. Then image restoration is done automatically, by filling these regions in with new information coming from the surrounding or cell in our case. In order to produce a perceptually plausible reconstruction, an inpainting technique must attempt to continue the isophotes (line of equal gray value) as smoothly as possible inside the reconstruction region. In other words the missing region should be inpainted so that inpainted gray value and gradient extrapolate the gray value and gradient outside the region. Several inpainting methods are based on the above ideas. Bertalmio et al. first introduced the notion of digital image inpainting and used third order partial differential equations (PDE) [4], [5], [9] to diffuse the known image information into the missing regions. Later, this inpainting approach was modified to take into account the direction of the level lines, called isophotes, and to relate it to the Navier-Stokes flow [6], [7, [21]. This operation propagates information into the masked region while preserving the edges. Anisotropic diffusion is used to preserve edges across the inpainted regions [13]. For further discussion of various methods, see the recent survey articles [10], [19]. The algorithms proposed in the literature differ depending on the assumptions made about the properties of the image. For example, the total variation (TV) inpainting model proposed, based on the Euler–Lagrange equation, employs anisotropic diffusion based on the contrast of the isophotes inside the inpainting domain [8]. This model, designed for inpainting small regions, does a good job at removing noise, but does not connect broken edges (single lines embedded in a uniform background). The Curvature- Driven Diffusion (CDD) model, extends the TV algorithm to also take into account geometric information of isophotes when defining the „strength‟ of the diffusion process, thus allowing the inpainting to proceed over larger areas. Although some of the broken edges are connected by the CDD approach, the resulting criteria for stopping the inpainting, the process is constantly applied to all masked pixels, regardless of the local smoothness of the region. As a result, computationally expensive operations might be unnecessarily performed, resulting in lengthy processing time. Thus, although non-linear PDE- based image restoration methods have the potential of