Fix figures and add intro figure
This commit is contained in:
1 parent
74109b2c4c
commit
cd670b8b51
51 files changed
+243
-395
No files matched your search
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
Executable → Regular
File mode changed.
@@ -1,177 +0,0 @@
|
||||
|
||||
|
||||
\newcommand{\scale}{1}
|
||||
|
||||
\newcommand{\frameScale}{.3}
|
||||
% \newcommand{\TikZxscale}{0.55}
|
||||
\newcommand{\xfshift}{8cm}
|
||||
% \newcommand{\TikZyslant}{0.4}
|
||||
% \newcommand{\TikZyfshift}{0.05cm}
|
||||
|
||||
\definecolor{colorFrameBackground}{rgb}{.82,.89,.95}
|
||||
% \definecolor{colorErrorsRegion}{rgb}{.163,.547,.757}
|
||||
\definecolor{colorErrors}{rgb}{.16,.40,.40}
|
||||
\definecolor{colorErrorsRegion}{rgb}{.209,.705,.740}
|
||||
|
||||
|
||||
\usetikzlibrary{patterns}
|
||||
|
||||
\makeatletter
|
||||
\pgfdeclareshape{test shape}
|
||||
{
|
||||
\inheritsavedanchors[from=rectangle]
|
||||
\inheritanchor[from=rectangle]{center}
|
||||
|
||||
\backgroundpath
|
||||
{
|
||||
\def\my@color{.}%
|
||||
\ifx\tikz@strokecolor\pgfutil@empty
|
||||
\else
|
||||
\let\my@color\tikz@strokecolor
|
||||
\fi
|
||||
\pgfsetfillpattern{north east lines}{\my@color}
|
||||
\pgfpathrectanglecorners{\pgfpoint{0}{0}}{\pgfpoint{1cm}{1cm}}
|
||||
\pgfusepath{fill}
|
||||
}
|
||||
}
|
||||
\makeatother
|
||||
|
||||
\newcommand{\FillFrame}[1]
|
||||
{
|
||||
\draw[fill=#1,fill opacity=1, very thick] (0,0) rectangle (8+1,8+1); % Opacity
|
||||
% \draw[thick] (0,0) rectangle (#2+1,#2+1);
|
||||
% \foreach \x in {1,...,#2}
|
||||
% \draw (\x,0) -- ++(0,#2+1);
|
||||
% \foreach \y in {1,...,#2}
|
||||
% \draw (0,\y) -- ++(#2+1,0);
|
||||
}
|
||||
|
||||
\newcommand{\DrawFrame}[1]
|
||||
{
|
||||
\draw[fill=#1,fill opacity=1, very thick] (0,0) rectangle (8+1,8+1); % Opacity
|
||||
% \draw[thick] (0,0) rectangle (#2+1,#2+1);
|
||||
% \foreach \x in {1,...,#2}
|
||||
% \draw (\x,0) -- ++(0,#2+1);
|
||||
% \foreach \y in {1,...,#2}
|
||||
% \draw (0,\y) -- ++(#2+1,0);
|
||||
}
|
||||
|
||||
\newcommand{\DrawBlock}[3]
|
||||
{
|
||||
\DrawBlockWithSize{#1}{#2}{#3}{1}{1}
|
||||
}
|
||||
\newcommand{\DrawBlockWithSize}[5]
|
||||
{
|
||||
\draw[draw=#1, fill=#1] (#2,#3) rectangle ++(#4 cm,#5 cm);
|
||||
}
|
||||
|
||||
|
||||
\begin{tikzpicture}[use Hobby shortcut,closed=true]
|
||||
|
||||
\tikzstyle{refmv_line}=[thick,draw=black]
|
||||
\tikzstyle{refmv}=[refmv_line,-latex,densely dashed]
|
||||
\tikzstyle{refmv_0}=[refmv,]
|
||||
\tikzstyle{refmv_1}=[refmv_0]
|
||||
|
||||
\tikzstyle{errorBlock}=[draw=colorErrors, fill=colorErrors]
|
||||
\tikzstyle{errorPropagation}=[draw=colorErrorsRegion, fill=colorErrorsRegion]
|
||||
|
||||
\coordinate (ErrorPosition) at (4,4);
|
||||
|
||||
|
||||
\def\errorPosX{3}
|
||||
\def\errorPosY{6}
|
||||
|
||||
\begin{scope}[scale=\frameScale*.95, yshift=2*\TikZyfshift, xshift=-2*\xfshift]
|
||||
\DrawFrame{colorFrameBackground}
|
||||
% \node at (12cm/2,-1.5cm) { $f_{t}$ };
|
||||
|
||||
\draw[errorBlock] (\errorPosX,\errorPosY) rectangle ++(1cm,1cm);
|
||||
|
||||
\foreach \x in {\errorPosX,...,7}
|
||||
\draw[errorPropagation] (\x+1,\errorPosY) rectangle ++(1 cm,1 cm);
|
||||
\foreach \y in {0.1,1,2,3,4,5}
|
||||
\foreach \x in {0.1,1,2,3,4,5,6,7,8}
|
||||
\draw[errorPropagation] (\x,\y) rectangle ++(1 cm,1 cm);
|
||||
|
||||
\draw (0,\errorPosY) -- ++(\errorPosX + 1,0) -- ++(0,1) -- ++(8-\errorPosX,0);
|
||||
|
||||
\end{scope}
|
||||
|
||||
\begin{scope}[scale=\frameScale, yshift=2*\TikZyfshift, xshift=0*\xfshift, yslant=\TikZyslant,xscale=\TikZxscale]
|
||||
\DrawFrame{colorFrameBackground}
|
||||
\node at (12cm/2,-1.5cm) { $f_{t}$ };
|
||||
|
||||
\draw[errorBlock] (\errorPosX,\errorPosY) rectangle ++(1cm,1cm);
|
||||
\end{scope}
|
||||
|
||||
\begin{scope}[scale=\frameScale, yshift=2*\TikZyfshift, xshift=1*\xfshift, yslant=\TikZyslant,xscale=\TikZxscale]
|
||||
\DrawFrame{colorFrameBackground}
|
||||
\node at (12cm/2,-1.5cm) { $f_{t+1}$ };
|
||||
% \begin{scope}[yshift=3.8cm, xshift=2.4cm]
|
||||
% \draw[erros] plot[scale=.3,smooth, tension=.7] coordinates {(0,0) (.5,3) (2.5,4) (5,3.5) (7.5,4) (8.5,3) (8.5,1) (10,-1.5) (3.5,0) (.5,-1.5) (0,0)};
|
||||
% \end{scope}
|
||||
|
||||
\draw[errorPropagation] (\errorPosX,\errorPosY) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX,\errorPosY -1) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +1,\errorPosY) rectangle ++(1cm,1cm);
|
||||
|
||||
\end{scope}
|
||||
|
||||
\begin{scope}[scale=\frameScale, yshift=2*\TikZyfshift, xshift=2*\xfshift, yslant=\TikZyslant,xscale=\TikZxscale]
|
||||
\DrawFrame{colorFrameBackground}
|
||||
\node at (12cm/2,-1.5cm) { $f_{t+2}$ };
|
||||
|
||||
\draw[errorPropagation] (\errorPosX,\errorPosY) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +1,\errorPosY) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +2,\errorPosY) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX,\errorPosY -1) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX+1,\errorPosY -1) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX+1,\errorPosY -2) rectangle ++(1cm,1cm);
|
||||
\end{scope}
|
||||
|
||||
\begin{scope}[scale=\frameScale, yshift=3*\TikZyfshift, xshift=3*\xfshift, yslant=\TikZyslant,xscale=\TikZxscale]
|
||||
\DrawFrame{colorFrameBackground}
|
||||
\node at (12cm/2,-1.5cm) { $f_{t+3}$ };
|
||||
% \begin{scope}[yshift=3.1cm, xshift=1.9cm]
|
||||
% \draw[erros] plot[scale=.5,smooth, tension=.7] coordinates {(0,0) (.5,3) (2.5,4) (5,3.5) (7.5,4) (8.5,3) (8.5,1) (10,-1.5) (3.5,0) (.5,-1.5) (0,0)};
|
||||
% \end{scope}
|
||||
|
||||
\draw[errorPropagation] (\errorPosX ,\errorPosY) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +1,\errorPosY) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +2,\errorPosY) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX ,\errorPosY -1) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +1,\errorPosY -1) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +2,\errorPosY -1) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +0,\errorPosY -2) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +1,\errorPosY -2) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +0,\errorPosY -3) rectangle ++(1cm,1cm);
|
||||
|
||||
\end{scope}
|
||||
|
||||
\begin{scope}[scale=\frameScale, yshift=4*\TikZyfshift, xshift=4*\xfshift, yslant=\TikZyslant,xscale=\TikZxscale]
|
||||
\DrawFrame{colorFrameBackground}
|
||||
\node at (12cm/2,-1.5cm) { $f_{t+4}$ };
|
||||
% \begin{scope}[yshift=3cm, xshift=1.8cm]
|
||||
% \draw[erros] plot[scale=.6,smooth, tension=.7] coordinates {(0,0) (.5,3) (2.5,4) (5,3.5) (7.5,4) (8.5,3) (8.5,1) (10,-1.5) (3.5,0) (.5,-1.5) (0,0)};
|
||||
% \end{scope}
|
||||
\draw[errorPropagation] (\errorPosX ,\errorPosY) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +1,\errorPosY) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +2,\errorPosY) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +3,\errorPosY) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX ,\errorPosY -1) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +1,\errorPosY -1) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +2,\errorPosY -1) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +3,\errorPosY -1) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +4,\errorPosY -1) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +0,\errorPosY -2) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +1,\errorPosY -2) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +2,\errorPosY -2) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +0,\errorPosY -3) rectangle ++(1cm,1cm);
|
||||
\draw[errorPropagation] (\errorPosX +2,\errorPosY -3) rectangle ++(1cm,1cm);
|
||||
|
||||
\end{scope}
|
||||
|
||||
\path[draw, -latex, thick] (-.5cm,-.5cm) -- ++(12.5cm,0) node[pos=1, yshift=-.2cm] {$t$};
|
||||
|
||||
\end{tikzpicture}
|
||||
@@ -71,18 +71,28 @@ Finally, concealment-aware techniques are used at the encoder, to either adapt t
|
||||
\label{Section22:Localisation}
|
||||
|
||||
|
||||
Localization techniques break the predictive coding loop in order to reduce the likelihood of errors affect large regions of the video stream, both in the spatial and temporal domains.
|
||||
As expected since localization techniques break the predictive coding, it reduces the efficiency of such methods. This leads to lower compression ratios and increasing bitrate.
|
||||
The basic principle of localization techniques is to break the predictive coding loop at certain predictive points, in order to reduce the likelihood of errors affecting large regions of the video signal, considering both the spatial and temporal domains.
|
||||
Since these techniques break the predictive coding, they have a negative impact in compression efficiency as expected.
|
||||
|
||||
\begin{figure}[tb]
|
||||
\centering
|
||||
\includegraphics[width=0.7\columnwidth]{\TransmissionResiliencefigs/ErrorPropagation_1.png}
|
||||
\caption{Examples of spatial and temporal error propagation and how it can be reduce using localisation techniques \cite{ErrorResilienceAug2005}.}
|
||||
\label{fig2:ErrorPropagation}
|
||||
\centering
|
||||
%\subfloat{\input{./022_ErrorResilience/Figures/LocalizationTechniquesSpatial.tikz}}\\
|
||||
%\subfloat{\input{./022_ErrorResilience/Figures/LocalizationTechniquesTemporal.tikz}}
|
||||
\input{./022_ErrorResilience/Figures/LocalizationTechniques_v2.tikz}
|
||||
\caption{Examples of spatial and temporal error propagation and how it can be reduced using localisation techniques \cite{ErrorResilienceAug2005}.}
|
||||
\end{figure}
|
||||
|
||||
This category of error resilience techniques are able to reduce the error propagation at two levels, i.e., temporal and spatial levels as shown in Figure~\ref{fig2:ErrorPropagation}.
|
||||
As shown in the figure spatial techniques break the dependencies within a single frame, as temporal error propagation can be reduce by selectively breaking the dependencies across different frames.
|
||||
%\begin{figure}[tb]
|
||||
% \centering
|
||||
% \includegraphics[width=0.7\columnwidth]{\TransmissionResiliencefigs/ErrorPropagation_1.png}
|
||||
% \caption{Examples of spatial and temporal error propagation and how it can be reduce using localisation techniques \cite{ErrorResilienceAug2005}.}
|
||||
% \label{fig2:ErrorPropagation}
|
||||
%\end{figure}
|
||||
|
||||
This group of error resilience techniques are able to reduce the error propagation at two levels, i.e., temporal and spatial levels.
|
||||
This is illustrated in in Figure~\ref{fig2:ErrorPropagation}, where one can observe that spatial techniques break the dependencies within a single frame, while temporal error propagation can be reduce by selectively breaking the dependencies across different frames.
|
||||
|
||||
\HERE
|
||||
|
||||
\subsection{Error resilient rate-distortion optimisation}
|
||||
|
||||
@@ -93,26 +103,26 @@ A common localization technique addresses the error resilience problem of the ra
|
||||
where $D$ is the distortion of all pixels, $R$ the required bitrate, and $\lambda$ is the Lagrangian multiplier.
|
||||
In \cite{ROPEJun2000,ROPE2007,ROPERefGenJan2010} a rate distortion optimisation for error resilience was presented.
|
||||
In order to take into consideration possible network losses a \ac{rope} of the end-to-end distortion is applied at the encoder-side.
|
||||
Let $f_n^i$ denote the original value of pixel $i$ in frame $n$, $\hat f_n^i$ denote its encoder reconstruction and $\tilde f_n^i$ the decoder reconstruction after error concealment if needed.
|
||||
Since data loss may occur in the transmission channel, $tilde f_n^i$ is considered a random variable at the encoder.
|
||||
Let $f_t^i$ denote the original value of pixel $i$ in frame at instant $t$, $\hat f_t^i$ denote its encoder reconstruction and $\tilde f_t^i$ the decoder reconstruction after error concealment if needed.
|
||||
Since data loss may occur in the transmission channel, $\tilde f_t^i$ is considered a random variable at the encoder.
|
||||
The expected mean square error of each pixel is given by:
|
||||
\begin{align}
|
||||
E\left\{d_n^i\right\} &= E\left\{ \left( f_n^i - \tilde f_n^i\right)^2 \right\} \nonumber \\
|
||||
&= (f_n^i)^2 - 2 f_n^i E\left\{ \tilde f_n^i \right\} + E\left\{ \left(\tilde f_n^i\right)^2 \right\}
|
||||
E\left\{d_t^i\right\} &= E\left\{ \left( f_t^i - \tilde f_t^i\right)^2 \right\} \nonumber \\
|
||||
&= (f_t^i)^2 - 2 f_t^i E\left\{ \tilde f_t^i \right\} + E\left\{ \left(\tilde f_t^i\right)^2 \right\}
|
||||
\end{align}
|
||||
Then to take into consideration the channel losses packet loss probability ($p$) must be used in the distortion measurement.
|
||||
This leads to the following for intra-coded pixels:
|
||||
\begin{align}\label{EQ2:E2EDIntra}
|
||||
E\left\{ \tilde f_n^i \right\} &= \left(1-p\right) \hat f_n^i + p E\left\{ \tilde f_{n-1}^i \right\} \nonumber \\
|
||||
E\left\{\left( \tilde f_n^i \right)^2 \right\} &= \left(1-p\right) \left( \hat f_n^i \right) + p E\left\{ \left( \tilde f_{n-1}^i \right)^2 \right\},
|
||||
E\left\{ \tilde f_t^i \right\} &= \left(1-p\right) \hat f_t^i + p E\left\{ \tilde f_{n-1}^i \right\} \nonumber \\
|
||||
E\left\{\left( \tilde f_t^i \right)^2 \right\} &= \left(1-p\right) \left( \hat f_t^i \right) + p E\left\{ \left( \tilde f_{n-1}^i \right)^2 \right\},
|
||||
\end{align}
|
||||
and inter-coded pixels:
|
||||
\begin{align}\label{EQ2:E2EDInter}
|
||||
E\left\{ \tilde f_n^i \right\} &= \left(1-p\right) \left( \hat e_n^i + E\left\{ \tilde f_{n-1}^{j} \right\} \right) + p E\left\{ \tilde f_{n-1}^i \right\} \nonumber \\
|
||||
E\left\{\left( \tilde f_n^i \right)^2 \right\} &= \left(1-p\right) \left( \left( \hat e_n^i \right)^2 + 2 \hat e_n^i E\left\{ \tilde f_{n-1}^{j} \right\} + E\left\{ \left( \tilde f_{n-1}^{j} \right)^2 \right\} \right) + p E\left\{ \left( \tilde f_{n-1}^i \right)^2 \right\}
|
||||
E\left\{ \tilde f_t^i \right\} &= \left(1-p\right) \left( \hat e_t^i + E\left\{ \tilde f_{n-1}^{j} \right\} \right) + p E\left\{ \tilde f_{n-1}^i \right\} \nonumber \\
|
||||
E\left\{\left( \tilde f_t^i \right)^2 \right\} &= \left(1-p\right) \left( \left( \hat e_t^i \right)^2 + 2 \hat e_t^i E\left\{ \tilde f_{n-1}^{j} \right\} + E\left\{ \left( \tilde f_{n-1}^{j} \right)^2 \right\} \right) + p E\left\{ \left( \tilde f_{n-1}^i \right)^2 \right\}
|
||||
\end{align}
|
||||
where the inter-coded pixel $i$ is predicted from pixel $j$ in the previous frame.
|
||||
The prediction error $e_n^i$, is quantized to the value $\hat e_n^i$, which is transmitted to the decoder along with the motion information.
|
||||
The prediction error $e_t^i$, is quantized to the value $\hat e_t^i$, which is transmitted to the decoder along with the motion information.
|
||||
Equations \ref{EQ2:E2EDIntra} and \ref{EQ2:E2EDInter} allow to measure the distortion taking into consideration the compression error but also the possible of frame loss and consequent \ac{ec} distortion.
|
||||
This technique was used in \cite{ROPEJun2000} to switch between intra and inter coding modes reducing the temporal dependencies, and, thus achieving superior error robustness in video transmission. In \cite{AdvancesROPEHE2642007} this technique was optimized to quarter pixel estimation in \ac{h264} standard. Moreover, in \cite{ROPEBurst2009} this method is extended for burst losses by introducing the probability of burst losses into the original end-to-end estimation. This shown to improve the quality under frame loss.
|
||||
In \cite{EndToEndDistortionConcealment2014} the distortion estimation assumes a temporal \ac{ec} at the decoder-side.
|
||||
@@ -155,17 +165,17 @@ In order to reduce error propagation one can employ leaky predictions \cite{Leak
|
||||
This leads to an exponential decay of error propagation, which leads to higher error robustness.
|
||||
In order to generate such reference frames the following can be applied
|
||||
\begin{equation}\label{EQ2:LeakyPredConstant}
|
||||
\check f_n^i = \alpha \hat f_n^i + (1- \alpha)C,
|
||||
\check f_t^i = \alpha \hat f_t^i + (1- \alpha)C,
|
||||
\end{equation}
|
||||
where $\check f_n^i$ is a reference frame pixel, $\alpha$ is the leak factor, and $C$ is a constant value.
|
||||
where $\check f_t^i$ is a reference frame pixel, $\alpha$ is the leak factor, and $C$ is a constant value.
|
||||
The use of leaky predictions has been used in the past for error control especially in scalable video coding \cite{LeakyPredScalable2002}, where $C$ in Equation~(\ref{EQ2:LeakyPredConstant}) is replace by a lower layer error-free reconstruction. In case of single layer coding, the use of a constant value has significant impact on the coding efficiency leading to higher bitrate as shown in \cite{LeakyPredConstant1973}.
|
||||
|
||||
In order to overcome the coding efficiency problem weighted prediction can be used with similar advantages. Since \ac{h264} motion compensation can be applied from two individual previously encoded frames and combined together to generate a single prediction for the current frame \cite{WeightedBiPred2005}. In the case of \ac{hevc}, a method was proposed to force the use of dual inter-frame prediction~\cite{DualFrameME2015} leading to lower error propagations.
|
||||
In case of leaky prediction instead of applying motion compensation is applied once, but with an interpolated frame obtained as follow:
|
||||
\begin{equation}\label{EQ2:LeakyPredDual}
|
||||
\check f_n^i = \alpha \hat f_n^i + (1- \alpha) \hat f_{n-1}^i.
|
||||
\check f_t^i = \alpha \hat f_t^i + (1- \alpha) \hat f_{n-1}^i.
|
||||
\end{equation}
|
||||
Using $\check f_n^i$ as reference frame for prediction significantly decrease error propagation if only one of the original coded frames is lost (i.e., $\hat f_{n}^i$ or $\hat f_{n-1}^i$).
|
||||
Using $\check f_t^i$ as reference frame for prediction significantly decrease error propagation if only one of the original coded frames is lost (i.e., $\hat f_{n}^i$ or $\hat f_{n-1}^i$).
|
||||
|
||||
\begin{figure}[tb]
|
||||
\centering
|
||||
@@ -182,7 +192,7 @@ In \cite{ROILeakyPred2009} in order to increase the error robustness with lower
|
||||
Figure \ref{fig2:ErrorPropagationROI} shows an example of error propagation within two different regions: \ac{roi} region and background.
|
||||
As show in the figure, since the \ac{roi} regions have predictions from regions in the background error propagation across regions occurs.
|
||||
Therefore, the method in \cite{ROILeakyPred2009} uses a weighted prediction for the \ac{roi} region obtained as follows: 1) a globally optimal prediction obtained by unrestricted motion estimation, 2) a prediction obtained by restricting motion estimation within the \ac{roi} region. This limits the impact on \ac{roi} from errors that affect background region. Moreover, since the leaky predictions are confined to a specific regions higher coding efficiency is achieved when compared with other methods.
|
||||
Equation (\ref{EQ2:LeakyPredDual}) results in a finite impulse response based weighting of reference frames, as previously weighted frames are not considered. In \cite{ROPERefGenJan2010} an infinitive impulse response based scheme is proposed to increase the decay in error propagation. To accomplish that, in Equation (\ref{EQ2:LeakyPredDual}) the use of the previous encoded frame ($\hat f_{n-1}^i.$) is replaced by the previous interpolated frame ($\check f_n^i$).
|
||||
Equation (\ref{EQ2:LeakyPredDual}) results in a finite impulse response based weighting of reference frames, as previously weighted frames are not considered. In \cite{ROPERefGenJan2010} an infinitive impulse response based scheme is proposed to increase the decay in error propagation. To accomplish that, in Equation (\ref{EQ2:LeakyPredDual}) the use of the previous encoded frame ($\hat f_{n-1}^i.$) is replaced by the previous interpolated frame ($\check f_t^i$).
|
||||
The idea of combining multiple predictions was also been used in the \ac{hevc} standard by combining it with end-to-end distortion to select the optimal predictions.
|
||||
|
||||
\subsection{Reference picture selection}
|
||||
|
||||
Reference in new issue
Block a user