Adaptive directional filter kernels for ringing and staircase artifact reduction

A major challenge in video processing is handling input material with completely different quality levels (e.g. WebTV and HDTV). For low-quality content especially coding artifacts should be strongly reduced to achieve an acceptable output quality. This paper focuses on the reduction of staircase an...

Full description

Saved in:
Bibliographic Details
Main Authors: Bruggemann, M., Richter, M., Schroder, H., Springer, P., Erdler, O.
Format: Conference Proceeding
Language:English
Subjects:
adaptive filter
artifact reduction
ringing
staircase
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A major challenge in video processing is handling input material with completely different quality levels (e.g. WebTV and HDTV). For low-quality content especially coding artifacts should be strongly reduced to achieve an acceptable output quality. This paper focuses on the reduction of staircase and ringing artifacts utilizing adaptive directional bilateral filter kernels. Binary maps are generated to discriminate areas that are used to control the filter strength. In this context a novel method for ringing artifact area discrimination is presented. The filter coefficients are adjusted to the local detail level and edge direction, utilizing Principal Components Analysis results. Low-pass filtering based on the proposed adaptive kernel results in a strong reduction of ringing and staircase artifacts while details can be well preserved.