REMOVAL OF NOISE AND

ARTIFACTS FROM BIOMEDICAL

SIGNAL
Structured Noise
• Noise is always presents in digital images during
image acquisition, coding, transmission, and
processing steps. Noise is very difficult to remove it
from the digital images without the prior knowledge
of noise model. That is why, review of noise models
are essential in the study of image denoising
techniques.
• Noise is always presents in digital images during
image acquisition, coding, transmission, and
processing steps. Noise is very difficult to remove it
from the digital images without the prior knowledge
of noise model.
• Structured Noise - when noise is periodic and Non
stationary.
Structure-borne Noise
• Structure-borne sound results from an impact on, or a vibration
against, a part of a building fabric resulting in sound being radiated
from an adjacent vibrating surface. A typical example of
structure-borne sound is footsteps on a floor which can be heard in a
room below.
• structured noise refers to the response to some physical phenomena
which is not originated at the reservoir, such as fluid segregation in
wellbore or pressure leak due to a brief opening of a valve. Such noisy
responses commonly appear to mix up with flow regimes, hindering
further reservoir flow analysis.
STRUCTURED NOISE DETECTION
• If this noise is stationary, it has fixed amplitude, frequency and
phase. Structured noise caused by interferences among
electronic components.
• An ideal noise-free pressure data is smooth and slow varying.
But, often they are affected by structured noise and
unstructured random noise. Thus, real-life pressure derivative
data are often a combination of noise-free signal, structured
noise, random noise. The smooth, slowly varying noise-free
signal can be regarded as the underlying trend of the observed
noise corrupted pressure derivative data and the structured
noises are randomly occurring deviations from the usual
smooth signal for a brief interval of time and show oscillatory
behavior.
• From the spectral perspective, the noise-free signal
lies on the lower range of the frequency spectrum,
where else the structured noises are in the
mid-range, and the random unstructured noises are
in the higher range.
• SSA (Singular Spectrum Analysis) is suitable for
decomposing pressure derivative data into additive
components that can be meaningfully categorized
into trends representing the noise-free signal, phase
and amplitude modulated oscillatory components
capturing the structured noise components and the
random noise components.
• However, three algorithmic questions remain
unanswered.
• (1) How do we decompose signals efficiently and
accurately so the structured noise is separated
from the trends? Poorly performed decomposition
may combine structured noise and trend in the
same component, hindering the noise removal
process.
• (2) How do we identify components that contain
only structured noise? There can be one or more
components for structured noise.
• (3) How do we temporally locate the structured
noise segments in the structured noise
components? The structured noise can be of
variable length and appear at an arbitrary time.
Structured Noise Component Identification
• Once the time series is decomposed into additive components, the
next objective is to identify the components that capture most of the
structured noise information.
• This component of interest identification step is formally called the
grouping.
• Similar to the embedding dimension selection step, there is no strict
rule about how to perform this grouping, rather heuristics are used
based on the type of the time series of interest and the purpose of
the analysis.
• An eigenvalue represents the amount of variance
captured by the corresponding component.
• In general, the largest eigenvalues are associated with
the trend components like the smooth, slow varying
noise-free signal, the intermediate ones are related
with the mid-frequency components like the
oscillatory structured noises, and lower values are
associated with the high-frequency random noises.
• And the structured noise components should be
projected onto the oscillating ones which has
mid-range eigenvalues as they account for the
short-duration oscillation around the trend.
Structured Noise Localization
• After SSA decomposition, the trend component
would represent the noise-free smooth signal, and
the structured noise component should have
non-zero values only in segments corresponding to
structured noises in the original signal.
• But often time due to high amplitudes of the
structured noises, component mixing occurs and they
affect the trend components.
• Also, in the structured noise components, noise oscillations expand to
neighboring regions, for example; where there are flow regimes in
between two structured noises, or flow regimes close to high
amplitude structured noises.
• Thus, to localize structured noise segments from the structured noise
components, we use a threshold value to compare against the
absolute value of the summation of the structured noise components,
and the sample data points exceeding the threshold value are
identified as structured noise
Physiological Noise
• Physiological noise deals with your own abilities to see and hear, your
state of health, whether you are tired or hungry at the time of the
communication, or any of many different physiological issues that can
interfere with paying attention to a message.
• While you cannot do much as a communicator to allay other
individuals’ physiological noise, you can pick up visual cues during
in-person, real-time communications and adjust your message
accordingly.
• For example, you can speak more slowly or loudly, or be more
succinct if you see your audience’s interest waning before lunch.
• For both in-person and electronic communications, you can offer
electronic versions of your information to audience members who
may need to increase font size.
• Be aware that physiological noise exists, and be prepared to adjust to
the communication situation and your audience’s needs.
 STATIONARY PROCESS
•A stationary process (or strict(ly) stationary process or strong(ly) stationary process) is
a stochastic process whose joint probability distribution does not change when shifted
in time or space.
•Consequently, parameters such as the mean and variance, if they exist, also do not
change over time or position.
•Stationarity is used as a tool in time series analysis, where the raw data are often
transformed to become stationary; for example, economic data are often seasonal
and/or dependent on a non-stationary price level.
•A "stationary process" is not the same thing as a "process with a stationary
distribution”
•Example a "time-homogeneous" Markov chain is sometimes said to have "stationary
transition probabilities". On the other hand, all stationary Markov random processes
are time-homogeneous.
 Weaker forms of stationarity:
Weak or wide-sense stationarity:
•A weaker form of stationarity commonly employed in signal processing is known
as weak-sense stationarity, wide-sense stationarity (WSS) or covariance
stationarity. WSS random processes only require that 1st and 2nd moments do
not vary with respect to time.
•Any strictly stationary process which has a mean and a covariance is also WSS.
 Second-order stationarity:
•It arises when the requirements of strict stationarity are only applied to pairs
of random variables from the time-series.
•It can be generalized to Nth order (for finite N) and strict stationary means
stationary of all orders.
•A process is second order stationary if the first and second order density
functions satisfy
ERGODIC PROCESS:
In signal processing, a stochastic process is said to be ergodic if its
statistical properties (such as its mean and variance) can be deduced
from a single, sufficiently long sample (realization) of the process.
STATIONARY ERGODIC PROCESS:
•A stochastic process which exhibits both stationarity and ergodicity.
•In essence this implies that the random process will not change its
statistical properties with time and that its statistical properties (such
as the theoretical mean and variance of the process) can be deduced
from a single, sufficiently long sample (realization) of the process.
•The theorem allows the time average of a conforming process to equal
the ensemble average.
Non-stationary noise
• Non-stationary noise, on the other hand, refers to noise that changes
significantly in its statistical properties over time.