Hierarchical fragmented image reassembly using a bundle-of-superpixel representation
We develop a new reassembly pipeline that automatically restores an image from its rips. First, we develop an image fragment descriptor called Bundle-of-Superpixel, which can more effectively support the detection of correlated (adjacent) image fragments and the computation of local pairwise partial matching. We also develop a hierarchical global composition algorithm that iteratively performs the reassembly through a bottom-up manner. We also build a data benchmark for fragmented images and propose a set of quantitative recomposition assessment measures. Through extensive experiments and comparisons, we demonstrate the effectiveness of our proposed algorithms.
Publication Source (Journal or Book title)
Computer Aided Geometric Design
Li, X., Xie, K., Hong, W., & Liu, C. (2019). Hierarchical fragmented image reassembly using a bundle-of-superpixel representation. Computer Aided Geometric Design, 71, 220-230. https://doi.org/10.1016/j.cagd.2019.04.015