Blockiness Distortion Measurement for Intra-Coded Video Frames by Using an Efficient NR Algorithm

Abstract

Lossy coding schemes are used in video encoding tools to fulfill the constraints of power and bandwidth. These coding schemes produce spatial and temporal distortions which affect the quality of the received video at user end. Blockiness is the most important and annoying distortion during compression. Measuring the video quality objectively enables dynamic monitoring and evaluation of emerging digital applications. In this work, Radon transform approach is used to propose a novel No-Reference (NR) blockiness metric. The proposed metric evaluates directional projection values along the block boundaries in both the horizontal and vertical directions and also takes into account the pixels near to the block boundaries. The difference of the directional projection values gives an estimate of blocking distortions. The proposed NR metric is simple, computationally efficient, and has a strong correlation with subjective data as proved by results.