A reduced-reference video structural similarity metric based on no-reference estimation of channel-induced distortion

Abstract

The reduced-reference (RR) approximation of a full-reference (FR) video quality assessment method is a convenient way to build evaluation metrics which are both intrinsically well correlated with human judgments and feasible to implement in a network scenario, without the need to explore the perceptual significance of new video features through mean opinion score tests. In this paper, we propose a RR approximation of the video structural similarity index (VSSIM), a FR metric which is known to be well descriptive of the video quality perceived by users. We focus on the visual degradation produced by channel transmission errors: first, at the encoder, a small set of salient structural video features is assembled and transmitted through the RR channel to the end-user; then, at the decoder the feature vector is combined with a fine-granularity, no-reference estimate of the channel-induced distortion to produce the VSSIM approximation. By uniformly quantizing the feature vector and compressing it using a context-adaptive, variable length encoder, we show that good correlation coefficients with ground-truth VSSIM (ρ = 0.85) may be achieved spending, respectively, less than 12 and 27 kbps for a video sequence with CIF or SD resolution.

Subjective evaluation of a NO-reference video quality Monitoring algorithm for H.264/AVC video over a noisy channel

“…b) NORM achieves almost the same performance as the full-reference PSNR, while working in no-reference mode. This was predictable, since we already showed in our earlier work [3] [4] that NORM estimates the true mean square error distortion very accurately, especially at the sequence level. Overall, we can claim that NORM can be effectively used to accurately predict the visual quality of H.264/AVC sequences transmitted over an error prone channel.…”

“…NORM produces such an estimate at the macroblock, frame and sequence level, by parsing the received H.264/AVC bitstream to extract information about coding modes, motion vectors and prediction residuals. In [4] the estimated MSE is fed forward in a reduced-reference quality monitoring scheme that computes an approximation of the Structural SIMilarity metrics (SSIM) [5], which typically shows a good correlation with the subjective Mean Opinion Score (MOS).…”

Reduced reference image and video quality assessments: review of methods

“…Low-bandwidth-based techniques are mostly used for video quality estimations and the values used in the performance are based on the amount of how much the multimedia content is distorted by compression technique. The higher the values of performance quality metrics (column-6), the higher is the quality of multimedia content and the better the technique for 9 High-bandwidth channel used for the video quality assessment using RR technique [131] Table 3 Review of the bitstream-based RR I and VQA metrics performance values with respect to the distortion types (JPEG, JPEG2000, etc.) used by authors in their RR technique to measure the quality metrics values (PSNR, SSIM, etc.…”

“…8. The RR methods that use low data for RR information are non-linear quantization [66] and distributed source coding [131]. RR methods based on high bandwidth can be either designed autonomously with respect to already existing FR methods [47] or as an approximation of some FR metrics as in [66].…”

Reduced-reference SSIM estimation

“…This scheme was further improved in [4] by making use of a divisive normalization transform (DNT). An RR video SSIM metric was proposed in [5] for quantifying visual degradations caused by channel transmission error. It is based on local spatial statistical features and uses distributed source coding techniques to reduce the required bandwidth to transmit RR features, though the resulting RR data rate is still much higher than those in [3] and [4].…”