Video coding technique with parametric modeling of noise

TL;DR

This paper introduces a novel parametric noise modeling technique for video encoding that improves subjective video quality when integrated with HEVC, validated through experiments showing a 1.8 MOS point enhancement.

Contribution

The paper proposes a new parametric noise modeling approach for video coding, demonstrating its effectiveness in improving subjective quality in practical encoding scenarios.

Findings

Noise modeling improves subjective video quality by ~1.8 MOS points.

Using noise reduction prior to encoding enhances video quality.

The method is validated with HEVC and real-time noise reduction.

Abstract

This paper presents a video encoding method in which noise is encoded using a novel parametric model representing spectral envelope and spatial distribution of energy. The proposed method has been experimentally assessed using video test sequences in a practical setup consisting of a simple, real-time noise reduction technique and High Efficiency Video Codec (HEVC). The attained results show that the use of the proposed parametric modeling of noise can improve the subjective quality of reconstructed video by approximately 1.8 Mean Opinion Scope (MOS) points (in 11-point scale) related to the classical video coding. Moreover, the present work confirms results attained in the previous works that the usage of even sole noise reduction prior to the encoding provides quality increase.