A study on image processing in DCT domain = DCT 영역에서 영상 처리에 대한 연구

With the development of compression standards such as JPEG and MPEC, which use Discrete Cosine Transform(DCT), most of digital image and video data are transmitted, and stored in the compressed bit-streams. Furthermore, the processing in DCT domain is smaller data volume and lower computation complexity than in spatial domain. It is resulted in growing an interest in technologies that can directly process image and video data in DCT domain. In this paper, we briefly introduce the several image manipulation methods, developing yet, in DCT domain, and propose two kind of new algorithms to directly process images in block-wise DCT domain. The first is to reduce blocking artifacts in image enhancement by using a new filtering algorithm. The second is to retrieve images efficiently by using global featuring and local featuring step by step. In experiment results, It is shown that the proposed algorithm of reducing blocking artifacts can significantly reduce the severe blocking artifacts occurred in filtering images in block-wise DCT. And, we also showed that the proposed algorithm of image retrieval has very available performance for retrieving images and very reasonable as considering human visual properties. Most of multimedia contents are circulated as the compressed format over DCT. In this viewpoints, the blocking artifacts reducing algorithm and the new image retrieval algorithm applied in block-wise DCT domain could be great foundation in advancement of communication, broadcast, and multimedia contents industries.
Ro, Yong-Manresearcher노용만researcher
Issue Date
392008/225023 / 000993881

학위논문(석사) - 한국정보통신대학원대학교 : 공학부, 2000, [ iv, 43 p. ]


필터링; 질감 검색; 블록 문제; DCT 변환; 영상처리; Texture indexing; Filtering; Blocking artifacts; Image Processing; DCT

Appears in Collection
School of Engineering-Theses_Master(공학부 석사논문)
Files in This Item
There are no files associated with this item.
  • Hit : 61
  • Download : 0
  • Cited 0 times in thomson ci


  • mendeley


rss_1.0 rss_2.0 atom_1.0