BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Asia/Seoul
X-LIC-LOCATION:Asia/Seoul
BEGIN:STANDARD
TZOFFSETFROM:+0900
TZOFFSETTO:+0900
TZNAME:KST
DTSTART:18871231T000000
DTSTART:19881009T020000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20230103T035312Z
LOCATION:Room 324\, Level 3\, West Wing
DTSTART;TZID=Asia/Seoul:20221209T140000
DTEND;TZID=Asia/Seoul:20221209T153000
UID:siggraphasia_SIGGRAPH Asia 2022_sess174_papers_484@linklings.com
SUMMARY:Constant Time Median Filter using 2D Wavelet Matrix
DESCRIPTION:Technical Communications, Technical Papers\n\nConstant Time Me
 dian Filter using 2D Wavelet Matrix\n\nMoroto, Umetani\n\nThe median filte
 r is a simple yet powerful noise reduction technique that is extensively a
 pplied in image, signal, and speech processing. It can effectively remove 
 impulsive noise while preserving the content of the image by taking the me
 dian of neighboring pixels; thus, it has various applications, such as res
 toration of a damaged image and facial beautification. The median filter i
 s typically implemented in one of two major approaches: the histogram-base
 d method, which requires O(1) computation time per pixel when focusing on 
 the kernel radius r, and the sorting-based method, which requires approxim
 ately O(r^2) computation time per pixel but has a light constant factor. T
 hese are used differently depending on the kernel radius and the number of
  bits in the image. However, the computation time is still slow, particula
 rly when the kernel radius is in the mid to large range. \n\nThis paper in
 troduces novel and efficient median filter with constant complexity O(1) f
 or kernel size using the wavelet matrix data structure, which has been app
 lied to query-based searches on one-dimensional data. We extended the orig
 inal wavelet matrix to two-dimensional data for application to computer gr
 aphics problems. The objective of this study was to achieve high-speed med
 ian filter computation in parallel computing environment with many threads
  (i.e., GPUs). Our implementation for the GPU is an order of magnitude fas
 ter than the histogram method for 8-bit images. Unlike traditional histogr
 am methods, which suffer from significant computational overhead, the prop
 osed method can handle images with high pixel depth (e.g., 16- and 32-bit 
 high dynamic range images). When the kernel radius is greater than 12 for 
 8-bit images, the proposed method outperforms the other median filter comp
 utation methods.\n\nRegistration Category: FULL ACCESS, ON-DEMAND ACCESS\n
 \nLanguage: ENGLISH\n\nFormat: IN-PERSON, ON-DEMAND
URL:https://sa2022.siggraph.org/en/full-program/?id=papers_484&sess=sess17
 4
END:VEVENT
END:VCALENDAR