Multi-level augmented inpainting network using spatial similarity

作者：

Highlights：

• A multi-level augmented inpainting network is proposed to harmonize the inter and intra level contexts rationally.

• A pyramid reconstruction structure is designed to realize the inter-level feature fusion.

• A spatial similarity based attention mechanism is proposed to guarantee the local intra-level continuity.

摘要

•A multi-level augmented inpainting network is proposed to harmonize the inter and intra level contexts rationally.•A pyramid reconstruction structure is designed to realize the inter-level feature fusion.•A spatial similarity based attention mechanism is proposed to guarantee the local intra-level continuity.

论文关键词：Image inpainting,Spatial information,Spatial similarity,Pyramid reconstruction structure

论文评审过程：Received 13 July 2021, Revised 20 November 2021, Accepted 21 January 2022, Available online 23 January 2022, Version of Record 3 February 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2022.108547