Large-scale Image Completion By Enhancing Structure Information

Image inpainting or image completion solves the problem of filling missing regions using known pixels around the region in a visually plausible way with consistent textures and continues structures. Image inpainting has been widely used in the field of image processing and computer vision. With the development of electronic technology and sensor technology in recent years, people now can make use of various types of equipment to get digital images, hence the demand of getting more precision and real-time algorithm for image inpainting is growing.In this paper we took repairing both texture and structure information of large-scale area for goal. Through a comprehensive analysis of several classical research in this field and a comparison of advantages and disadvantages in details among these methods, we proposed the shortcomings and challenges in the field of image inpainting. Then in order to overcome the shortcomings of artifacts in structure component, we thought from two different aspects and make two main contributions,1. We first use statistical feature for block matching and use user interaction to enhance structure component. We propose a novel patch-based inpainting algorithm combining block matching feature and Randomized Correspondence Algorithm. This method can overcome the problem of low efficiency and low quality for greedy-based inpainting methods. Experimental results after comparing with various optimized patch-based inpainting methods show the effectiveness of our method.2. Then we combine structure description feature from computer vision and Graph-Cut energy optimization methods to design a novel inpainting algorithm, which will not only remove error accumulation phenomenon of patch-based methods, but also make complexity structure feature repaired consistently. Experiments on a wide variety of images show our method yields better results in various challenging cases than state-of-art methods both on visual impact and efficiency.Our research shows that by increasing the importance of structure information in the repair procedure, more consistent result can be got since human eyes will be more sensitive to structure components. Besides, combining high-efficient matching technology, the two algorithms in this paper can guarantee the algorithm running efficiently on the basis of good performance especially for complexity structures.