Image Super-Resolution using RIPOC-based registration
-
- Taniguchi Kazuki
- Ritsumeikan University of Information Science and Engineering
-
- Han Xian-Hua
- Ritsumeikan University of Information Science and Engineering
-
- Iwamoto Yutaro
- Ritsumeikan University of Information Science and Engineering
-
- Sasatani So
- Ritsumeikan University of Information Science and Engineering
-
- Ohashi Motonori
- Ritsumeikan University of Information Science and Engineering
-
- Chen Yen-Wei
- Ritsumeikan University of Information Science and Engineering
Bibliographic Information
- Other Title
-
- 回転不変位相限定相関法の超解像技術への応用(インタラクティブシステム・画像入力デバイス・方式,及び一般)
- 回転不変位相限定相関法の超解像技術への応用
- カイテン フヘン イソウ ゲンテイ ソウカンホウ ノ チョウカイゾウ ギジュツ エ ノ オウヨウ
Search this article
Abstract
Multi-Frame Super-Resolution is a technique which attempts to reconstruct high-resolution image by fusing a number of low-resolution images. Therein, high-accuracy registration for the input low-resolution image set is required for reconstruction a clear high-resolution image. The conventional registration approaches for Multi-Frame Super-Resolution such as Iterative Back Projection and LK optical flow methods, can obtain accurate parameters for tiny translation or rotation movement between images. However, for large movement especially in rotation, there are still no efficient methods for reconstructing high-resolution image. Therefore, in this paper we proposed a new method using Rotation Invariant Phase Only Correlation, which can estimate large rotation parameter, and then follow the Iterative back Projection method for reconstructing high-resolution image. Experimental results show that the reconstructed high resolution images by our proposed approach are much better than those by interpolation techniques.
Journal
-
- ITE Technical Report
-
ITE Technical Report 35.23 (0), 33-36, 2011
The Institute of Image Information and Television Engineers
- Tweet
Details 詳細情報について
-
- CRID
- 1390001204529366784
-
- NII Article ID
- 110008687538
-
- NII Book ID
- AN1059086X
-
- ISSN
- 24241970
- 13426893
-
- NDL BIB ID
- 11178277
-
- Text Lang
- ja
-
- Data Source
-
- JaLC
- NDL
- CiNii Articles
-
- Abstract License Flag
- Disallowed