Efficient and effective image enlargement without distortion attracts numerous interests these days. This paper presents an image enlargement system for images with repetitive components. Based on repetitive components and seam filling, our system achieves distortion-free visual effect. Our system captures and extracts the repetitive components by MSERs. After which, we got the repetitive component (RC) regions and non-repetitive component (NRC) regions. In order to deal with the different characteristics of these two regions, we intuitively incorporated two different algorithms: patch-based texture synthesis for RC regions and in painting for NRC regions. To achieve seamlessness filling, we lastly smooth the boundary between the filled regions and its surroundings by multi-resolution. Our system shows great success in enlarging image with repetitive components. Copyright © 2014 IEEE.
|Title of host publication||Proceedings: 2014 International Conference on Virtual Reality and Visualization, ICVRV 2014|
|Editors||Xukun SHEN, Xiaopeng ZHANG, Zhong ZHOU, Guodong ZHANG, Xun LUO|
|Place of Publication||Piscataway|
|Publisher||Institute of Electrical and Electronics Engineers, Inc|
|Publication status||Published - 2014|