Non-binary circle hough transform on GPU architecture그래픽스 처리장치 아키텍쳐에서의 비이진 원형 허프 변환

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 437
  • Download : 0
Array conversion method is a favorable approach to utilize the Graphics Processing Unit (GPU) in Hough Circle Detection. However, applying the method on non-binary images needs binarization as a preprocess which causes much information loss. To avoid the loss, an approach that scans entire pixel data adaptively to the GPU architecture is needed, without binarization. In this paper, an advanced Hough transform algorithm on non-binary images was implemented considering memory-thread unity and memory locality to be adapt to the GPU architecture using low-cost operations. Results showed that the proposed method on non-binary images achieved 20% improved accuracy maintaining the performance and the power efficiency of the array conversion and conventional method on binary images nearly matched. The proposed method would be a desirable choice on real-time applications that require more accurate non-binary Hough circle detection.
Advisors
Han, Dongsooresearcher한동수researcher
Description
한국과학기술원 :전산학부,
Publisher
한국과학기술원
Issue Date
2019
Identifier
325007
Language
eng
Description

학위논문(석사) - 한국과학기술원 : 전산학부, 2019.2,[iv, 23 p. :]

Keywords

Hough transform▼agraphics processing unit▼aparallel algorithm▼anon-binary images▼acomputer vision; 허프 변환▼a그래픽스 처리장치▼a병렬 알고리즘▼a비이진 이미지▼a컴퓨터 영상

URI
http://hdl.handle.net/10203/267114
Link
http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=843518&flag=dissertation
Appears in Collection
CS-Theses_Master(석사논문)
Files in This Item
There are no files associated with this item.

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0