Paper Abstract and Keywords |
Presentation |
2015-11-12 13:00
Local image inpainting with high frequency resolution analysis Yuya Inuzuka, Kazuki Sakashita, Masaya Hasegawa, Shigeki Hirobayashi, Tadanobu Misawa (Toyama Univ.), Hiroyuki Shioya (Muroran Institute of Tech.) |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
We present a new technique for image restoration. Although several algorithms can modify images in an undetectable manner for object removal or image recovery, the recovered spectral structure differs from the original one. In this research, we obtain a recovered spectral structure resembling the original one from a known region surrounding a missing region. The proposed method is based on a previously proposed concept, known as non-harmonic analysis (NHA). NHA can be used to extract accurate spectra, irrespective of the window function, and frequency resolution of low frequencies is higher than that of the discrete Fourier transform. We conducted a numerical experiment to verify the accuracy of the spectrum extracted from an artificial signal. The results carried out on several artificially produced damaged images show that a high-quality image is obtained. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Image restoration / Object removal / Non-harmonic analysis / / / / / |
Reference Info. |
ITE Tech. Rep. |
Paper # |
|
Date of Issue |
|
ISSN |
|
Download PDF |
|