Efficient Region-of-Interest Scalable Video Coding with Adaptive Bit-Rate Control

Abstract

This work relates to the regions-of-interest (ROI) coding that is a desirable feature in future applications based on the scalable video coding, which is an extension of the H.264/MPEG-4 AVC standard. Due to the dramatic technological progress, there is a plurality of heterogeneous devices, which can be used for viewing a variety of video content. Devices such as smartphones and tablets are mostly resource-limited devices, which make it difficult to display high-quality content. Usually, the displayed video content contains one or more ROI(s), which should be adaptively selected from the preencoded scalable video bitstream. Thus, an efficient scalable ROI video coding scheme is proposed in this work, thereby enabling the extraction of the desired regions-of-interest and the adaptive setting of the desirable ROI location, size, and resolution. In addition, an adaptive bit-rate control is provided for the region-of-interest scalable video coding. The performance of the presented techniques is demonstrated and compared with the joint scalable video model reference software (JSVM 9.19), thereby showing significant bit-rate savings as a tradeoff for the relatively low PSNR degradation.

Complexity-Aware Adaptive Preprocessing Scheme for Region-of-Interest Spatial Scalable Video Coding

“…It should be noted that a minor decrease in the ROI quality could be expected due to reducing the background size in favor of the transition region, which has the gradually increased quality toward the ROI. This is the reason why PSNR values of plots regarding the ROI quality without any transition region (which relate to the authors' previous research [32], [33]) are sometimes higher than those with the varying number of prefiltered straps (especially, comparing the larger number of prefiltered straps because of the more gradual quality transition toward the background).…”

“…The prefilters, which are used in the proposed novel scheme of Fig. 1 are Gaussian filters due to their relatively low computational complexity, as explained for example, in the authors previous research [32], [33], which present a detailed analysis of Gaussian, Wiener, and Wavelet filters as well as an analysis of various filter settings. However, [32], [33] do not introduce an adaptive transition region to be provided between the ROI and background for significantly improving visual presentation quality, similarly to [34], [35], which proposes to prefilter an entire frame according to varying network conditions (e.g., the network bandwidth change and packet loss ratio).…”

“…2. It is noted that plots, which relate to the ROI quality without any transition region [i.e., defined as PSNR of ROI (no transition region)] are based on authors' previous research [32], [33]. the standard deviation (σ) of each prefiltered strap gradually increases by 0.1 up to σ = 1.5.…”

Low Complexity Encoder with Multilabel Classification and Image Captioning燤odel

“…One of the drawbacks is that the computational complexity is enhanced. Grois and Hadar [15] projected coding for ROI, which is decided from preencoded scalable video bitstream and adaptive settings activate the extraction of required ROI by position, size, and resolution. It offers an effective approach for meeting satisfy the demand for dissimilar end-user devices.…”

Towards a Low Complexity Scheme for Medical Images in Scalable Video Coding

“…The main drawback of this work is the increase in complexity. Grois and Hadar [19] proposed coding for regions of interest, which were flexibly selected from the pre-encoded scalable video bit stream and that with adaptive settings enabled the extraction of desirable regions of interest by location, size, and resolution. Their methodology provides an efficient mechanism to meet the requirement of heterogeneous end-user devices.…”