Presentation | 2023-03-02 Blind deconvolution with non-smooth regularization via Bregman proximal DC algorithms Shota Takahashi, Mirai Tanaka, Ikeda Shiro, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | Blind deconvolution is a technique to recover an original signal without knowing a convolving filter from its convolution. It is formulated as a minimization problem of a quartic loss function under some assumption. In this paper, we find a difference of convex functions (DC) decomposition for the loss function and apply the Bregman proximal DC algorithm (BPDCA) and the BPDCA with extrapolation (BPDCAe), which is an acceleration of BPDCA. Using this DC decomposition, we obtain the $L$-smooth adaptable ($L$-smad) property, which guarantees the global convergence of BPDCA. When our regularizer has a sufficiently simple structure, the subproblem of BPDCA(e) at each iteration is solved in a closed-form expression, and thus our algorithms solve large-scale problems efficiently. We also provide the stability analysis around the equilibrium point and demonstrate our proposed algorithms through numerical experiments on image deblurring. The results show that BPDCAe successfully recovered the original image and outperformed other existing algorithms. This paper is based on [21]. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Blind deconvolution / DC optimization / Bregman proximal DC algorithms / Image deblurring |
Paper # | PRMU2022-79,IBISML2022-86 |
Date of Issue | 2023-02-23 (PRMU, IBISML) |
Conference Information | |
Committee | PRMU / IBISML / IPSJ-CVIM |
---|---|
Conference Date | 2023/3/2(2days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Future University Hakodate |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | Seiichi Uchida(Kyushu Univ.) / Masashi Sugiyama(Univ. of Tokyo) |
Vice Chair | Takuya Funatomi(NAIST) / Mitsuru Anpai(Denso IT Lab.) / Toshihiro Kamishima(AIST) / Koji Tsuda(Univ. of Tokyo) |
Secretary | Takuya Funatomi(CyberAgent) / Mitsuru Anpai(Univ. of Tokyo) / Toshihiro Kamishima(NTT) / Koji Tsuda(Hokkaido Univ.) |
Assistant | Nakamasa Inoue(Tokyo Inst. of Tech.) / Yasutomo Kawanishi(Riken) / Yoshinobu Kawahara(Osaka Univ.) / Taiji Suzuki(Tokyo Inst. of Tech.) |
Paper Information | |
Registration To | Technical Committee on Pattern Recognition and Media Understanding / Technical Committee on Information-Based Induction Sciences and Machine Learning / Special Interest Group on Computer Vision and Image Media |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Blind deconvolution with non-smooth regularization via Bregman proximal DC algorithms |
Sub Title (in English) | |
Keyword(1) | Blind deconvolution |
Keyword(2) | DC optimization |
Keyword(3) | Bregman proximal DC algorithms |
Keyword(4) | Image deblurring |
1st Author's Name | Shota Takahashi |
1st Author's Affiliation | The Graduate University for Advanced Studies(SOKENDAI) |
2nd Author's Name | Mirai Tanaka |
2nd Author's Affiliation | The Institute of Statistical Mathematics(ISM) |
3rd Author's Name | Ikeda Shiro |
3rd Author's Affiliation | The Institute of Statistical Mathematics(ISM) |
Date | 2023-03-02 |
Paper # | PRMU2022-79,IBISML2022-86 |
Volume (vol) | vol.122 |
Number (no) | PRMU-404,IBISML-405 |
Page | pp.pp.111-118(PRMU), pp.111-118(IBISML), |
#Pages | 8 |
Date of Issue | 2023-02-23 (PRMU, IBISML) |