Object removal and loss concealment using neighbor embedding methods
作者：
Highlights：
• Exemplar-based image inpainting is revisited as a neighbor embedding problem.
• An improved K-NN search method using linear regression based subspace mappings is proposed.
• A new priority order, the advantage of which is shown for object removal, is introduced.
• The performances of the resulting inpainting algorithms are assessed in two application contexts: loss concealment and object removal.
摘要
Highlights•Exemplar-based image inpainting is revisited as a neighbor embedding problem.•An improved K-NN search method using linear regression based subspace mappings is proposed.•A new priority order, the advantage of which is shown for object removal, is introduced.•The performances of the resulting inpainting algorithms are assessed in two application contexts: loss concealment and object removal.
论文关键词：Image inpainting,Neighbor embedding,Least Squares approximation
论文评审过程：Received 30 August 2012, Revised 26 June 2013, Accepted 27 August 2013, Available online 12 September 2013.
论文官网地址：https://doi.org/10.1016/j.image.2013.08.020
