Project Report for Image Quilting Reproduction

Abstract

The demand for high-quality images has been greater in computer graphics and computer vision. Image-based rendering is playing bigger roles in video games, for instance. Rather than creating the whole physical world from scratch, we prefer to pick samples from small images and synthesize realistic novel views. The advantage of texture synthesis is that it is faster and requires a smaller input dataset. Also, texture transfer is a useful tool in computer graphics. Texture synthesis is the process of creating an image of arbitrary size from a different image texture. Both texture mapping work directly on flat images and require no three dimensional information [2]. In this project we implement the image quilting algorithm for texture synthesis and texture transfer, proposed in SIGGRAPH 2001 paper by Alexei A. Efros and William T. Freeman. The main idea is to compose a large patch of texture by “stitching together” arbitrarily shaped smaller patches from the input texture, that is, image quilting. Texture transfer would be implemented based on texture synthesis