A fast median filter using AltiVec

  title={A fast median filter using AltiVec},
  author={Priyadarshan Kolte and Roger Smith and Su Wen},
  journal={Proceedings 1999 IEEE International Conference on Computer Design: VLSI in Computers and Processors (Cat. No.99CB37040)},
  • Priyadarshan KolteRoger SmithS. Wen
  • Published 10 October 1999
  • Computer Science
  • Proceedings 1999 IEEE International Conference on Computer Design: VLSI in Computers and Processors (Cat. No.99CB37040)
The paper describes the design and implementation of a median filter for graphics images on the Motorola AltiVec architecture. [] Key Method The median filter is based on a new sorting network which sorts N/sup 2/ numbers (arranged in an N/spl times/N square) by sorting all columns, rows, and diagonal lines in the square. The paper also describes a scheme for efficient testing of the sorting network.

Figures from this paper

Bidimensional median filter for parallel computing architectures

  • R. M. SanchezP. Rodríguez
  • Computer Science
    2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
  • 2012
A novel sorting algorithm, with O(n) computational complexity and a highly parallelizable structure, based on the Complementary Cumulative Distribution Function is presented and a 2D median filter based on this proposed sorting algorithm can achieve O(1) complexity.

Highly Parallelable Bidimensional Median Filter for Modern Parallel Programming Models

This paper greatly improves the results of earlier work, and proposes a trade-off between O(1) and $O(n) complexity in order to improve the overall throughput; additionally, it makes use of the Salt & Pepper noise model to improved the image reconstruction quality with a small performance impact.

Scalable block processing algorithms

This thesis presents the theory and design methods required to design a block processing algorithm with linear scalability for any stream processing system and shows how these methods can benefit the implementation of various specific systems like the moving sum, recursive filters and rank order filters.

A Parallel Memory Suitable for SIMD Architecture Processing High-Definition Image Haze Removal in High-Speed

A parallel memory model which is suitable for SIMD architecture processing haze removal on high-definition images in high-speed is proposed.

Multi-Dimensional Parallel Rank Order Filtering

This work presents a method to design multi-dimensional rank order filters based on sorting networks composed from Batcher¿s merging networks and introduces an additional optimization that increases the savings obtained by pruning sorting networks.

An Implementation and Improvement of Fast Two-Dimensional Median Filtering

An improved way -- Quasi-median filtering algorithm -- have been proposed to reduce the occupancy rate of FPGA resources on the premise of ensuring the result of median filtering, and conclusions can be drawn that such improvements can achieve better filtering results, and can reduceFPGA resource utilization.

Dense stereo disparity maps by Dynamic Time Warp with sparse features and FPGA acceleration

  • Zheng QianK. Takaya
  • Computer Science
    2011 24th Canadian Conference on Electrical and Computer Engineering(CCECE)
  • 2011
A Dynamic Time Warp (DTW) algorithm is proposed that provides excellent results of dense stereo disparity maps in accuracy and an improvement on the DTW algorithm is suggested to reduce the computational expense so as to meet the requirement of real-time processing.

Subword sorting with versatile permutation instructions

  • Z. ShiR. Lee
  • Computer Science
    Proceedings. IEEE International Conference on Computer Design: VLSI in Computers and Processors
  • 2002
This paper shows how one of the bit permutation instructions, GRP, can be used for fast sorting and demonstrates the versatility of this permutation instruction for uses other than bit permutations.


A novel vector register file design (VRF) that extends the capability of a conventional VRF design that can be designed readily in VLSI technology with a little overhead, while reserving the original structure of the VRF.

Motion-tracking adaptive persistence and adaptive-size median filter for color Doppler processing in ultrasound systems on multicore platform

Color Doppler processing in the ultrasound imaging systems is mainly used to observe the blood flow in the region of interest. The desired blood signal will be greatly affected by the speckle noises,



Networks for Sorting

AltiVec Vectorizes PowerPC: 5/11/98

AltiVec (formerly known as VMX) takes a step beyond all other multimedia extensions by using 128-bit registers and ALUs, twice the width of competing designs, and provides other advantages over MMX and similar extensions, such as Sun’s VIS and MDMX for MIPS.

Introduction to Parallel Algorithms and Architectures: Arrays, Trees, Hypercubes

This chapter discusses sorting on a Linear Array with a Systolic and Semisystolic Model of Computation, which automates the very labor-intensive and therefore time-heavy and expensive process of manually sorting arrays.

Paeth, \Median Finding on a 3 3 Grid," in Graphics Gems

  • 1990

Median finding on a 3 x 3 grid

scheduling," Advanced Compiler Design and Implementation

  • 1997