IMAGE ARTIFACTS AND NO-REFERENCE FEATURE EXTRACTIONS: A REVIEW

Abstract

Image attributes, such as shape, orientation, size, texture, gradient, luminance, brightness, and
contrast can be easily recognized by human visual system. All these attributes contribute to the
characteristics of an image or image features, which reveal uniqueness of the image. Once the
images get corrupted due to compression or transmission, the unexpected artificial features, which
are not existing in the origi"al images are introduced. The characteristics of these unexpected
features known as artifact characteristics vary from one artifact to another By knowing the original
image features, un-expected features can be detected. There are a number of feature extraction
techniques that can be used to extract image features, in order to observe the presence of artifacts.
These techniques can be generally classified into two categories, as spatial domain based and
transform domain based. This paper presents a review of the existing spatial domain artifact
extraction techniques using no-reference (withoul comparison to original image) as they are
computationally inexpensive and suitable for in-service quality monitoring.
Keywords: Artifact, feature extraction, error, blocking, blur, ringing, masking, lost blocks.