Title :
Elimination of bias in the FIR-median hybrid filter
Author :
Longbotham, H.G. ; Zhang, M. ; Engelken, E.
Author_Institution :
Nonlinear Signal Process. Lab., Texas Univ., San Antonio, TX, USA
Abstract :
In Heinonen and Neuvo (1988), the FIR-median hybrid (FMH) filter, a class of median type filters for image processing was proposed. In these filters, linear FIR predictors are used in conjunction with the median operation. In the present paper, the one dimensional FMH is shown to be biased at boundaries and does not smooth junctions well for one dimensional signals. In addition it does not preserve corners of less than 900 or line endings for two dimensional signal. The residual technique introduced can solve both of these problems
Keywords :
FIR filters; image processing; median filters; prediction theory; two-dimensional digital filters; FIR-median hybrid filter; FMH filter; bias; boundaries; corners; image processing; junctions; line endings; linear FIR predictors; median type filters; one dimensional FMH filter; one dimensional signal; residual technique; two dimensional signal; Concatenated codes; Finite impulse response filter; Focusing; Image processing; Nonlinear filters; Signal processing;
Conference_Titel :
Image Processing, 1994. Proceedings. ICIP-94., IEEE International Conference
Conference_Location :
Austin, TX
Print_ISBN :
0-8186-6952-7
DOI :
10.1109/ICIP.1994.413424
