Presentation 2020-01-23
[Invited Talk] Accelerating Image Processing by Constant-time Filtering
Kenjiro Sugimoto,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Digital filter plays a fundamental role in a variety of methods in image processing, computer vision and computer graphics. As many modern methods such as deep convolutional neural networks and convex optimization do, it is common to apply filters with various parameter values to an image or a set of images. In this situation, reducing computational complexity of image filters is a must to achieve practical processing performance. A potential solution to this problem is textit{constant-time} filter where constant-time, which is often abbreviated to $O(1)$, means that per-pixel computational complexity is independent of filter window size, i.e., an $N$-pixel image can be filtered in $O(N)$ time as a whole. This theoretical property is particularly preferred in high-resolutional or high-dimensional image processing since it tends to require wider window area. Hence, it is strongly demanded to establish efficient and advanced constant-time algorithms. This talk summarizes constant-time filters the author have researched in this decade which mixing their background at that time.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Constant-time filtering / Gaussian filtering / bilateral filtering
Paper # IT2019-59,SIP2019-72,RCS2019-289
Date of Issue 2020-01-16 (IT, SIP, RCS)

Conference Information
Committee IT / SIP / RCS
Conference Date 2020/1/23(2days)
Place (in Japanese) (See Japanese page)
Place (in English) Hiroshima City Youth Center
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair Jun Muramatsu(NTT) / Naoyuki Aikawa(TUS) / Tomoaki Otsuki(Keio Univ.)
Vice Chair Tadashi Wadayama(Nagoya Inst. of Tech.) / Kazunori Hayashi(Osaka City Univ) / Yukihiro Bandou(NTT) / Satoshi Suyama(NTT DoCoMo) / Fumiaki Maehara(Waseda Univ.) / Toshihiko Nishimura(Hokkaido Univ.)
Secretary Tadashi Wadayama(Saga Univ.) / Kazunori Hayashi(Nagano Pref Inst. of Tech.) / Yukihiro Bandou(Hiroshima Univ.) / Satoshi Suyama(Hosei Univ.) / Fumiaki Maehara(NTT) / Toshihiko Nishimura(Kyushu Univ.)
Assistant Hideki Yagi(UEC) / Kenjiro Sugimoto(Waseda Univ.) / Kazushi Muraoka(NEC) / Shinsuke Ibi(Doshisha Univ.) / Koichi Adachi(Univ. of Electro-Comm.) / Osamu Nakamura(Sharp) / Manabu Sakai(Mitsubishi Electric)

Paper Information
Registration To Technical Committee on Information Theory / Technical Committee on Signal Processing / Technical Committee on Radio Communication Systems
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) [Invited Talk] Accelerating Image Processing by Constant-time Filtering
Sub Title (in English)
Keyword(1) Constant-time filtering
Keyword(2) Gaussian filtering
Keyword(3) bilateral filtering
1st Author's Name Kenjiro Sugimoto
1st Author's Affiliation Waseda University(Waseda Univ.)
Date 2020-01-23
Paper # IT2019-59,SIP2019-72,RCS2019-289
Volume (vol) vol.119
Number (no) IT-376,SIP-377,RCS-378
Page pp.pp.133-133(IT), pp.133-133(SIP), pp.133-133(RCS),
#Pages 1
Date of Issue 2020-01-16 (IT, SIP, RCS)