A structure first image inpainting approach based on self-organizing map (SOM)

• Main Authors: CHEN, Bo, WANG, Zhaoxia, BAI, Ming, WANG, Quan, SUN, Zhen
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2010
• Subjects: Curve fitting; Image inpainting; Neural networks; Self-organizing map (SOM); Numerical Analysis and Scientific Computing; Operations Research, Systems Engineering and Industrial Engineering
• Online Access: https://ink.library.smu.edu.sg/sis_research/5556; https://ink.library.smu.edu.sg/context/sis_research/article/6559/viewcontent/2010_Structure_first_Image_Inpainting_Approach_av.pdf
• Institution: Singapore Management University

This paper presents a structure first image inpainting method based on self-organizing map (SOM). SOM is employed to find the useful structure information of the damaged image. The useful structure information which includes relevant edges of the image is used to simulate the structure information of the lost or damaged area in the image. The structure information is described by distinct or indistinct curves in an image in this paper. The obtained target curves separate the damaged area of the image into several parts. As soon as each part of the damaged image is restored respectively, the damaged image is inpainted. The efficiency of the proposed method is demonstrated by the simulation results.