Scribd
Upload
190 views

KALMAN FILTER Applications in Image Processing

The Kalman filter is a recursive algorithm that estimates the state of a linear dynamic system from a series of noisy measurements over time. It has applications in image processing by modeling the image as a linear system corrupted by noise. The filter uses a predictor-corrector approach, predicting the next state using a time update and correcting it using a measurement update that incorporates new measurements. An extended Kalman filter handles nonlinear systems through linearization. Complex Kalman filtering represents images using complex numbers.

Uploaded by

Hayder Kadhim
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
190 views

KALMAN FILTER Applications in Image Processing

The Kalman filter is a recursive algorithm that estimates the state of a linear dynamic system from a series of noisy measurements over time. It has applications in image processing by modeling the image as a linear system corrupted by noise. The filter uses a predictor-corrector approach, predicting the next state using a time update and correcting it using a measurement update that incorporates new measurements. An extended Kalman filter handles nonlinear systems through linearization. Complex Kalman filtering represents images using complex numbers.

Uploaded by

Hayder Kadhim
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 24

KALMAN FILTER

Applications in Image processing


Introduction
• The kalman filter is a recursive state space model based
estimation algorithm.

• This filter is named after Rudolph E. Kalman, who in


1960 published his famous paper describing a recursive
solution to the discrete data linear filtering problem
(Kalman 1960).

• This algorithm was basically developed for single


dimensional and real valued signals which are
associated with the linear systems assuming the system
is corrupted with linear additive white Gaussian noise.
• The Kalman filter addresses the general problem of
trying to estimate the state x ∈ ℜn of a discrete-time
controlled process that is governed by the linear
difference equation

xk = Axk – 1 + Buk – 1 + wk – 1

• with a measurement z that is

zk = Hxk + vk

• The random variables wk and vk represent the process


noise and measurement noise respectively.
• The nxn matrix A in the previous difference
equation relates the state at the previous time
step k-1 to the state at the current step k , in the
absence of either a driving function or process
noise.

• The nxl matrix B relates the optional control input


u to the state x.

• The mxn matrix H in the measurement equation


relates the state to the measurement zk.
The Computational Origins of the Filter :

• We define𝑥𝑘− ∈ℜn to be our a priori state


estimate at step k given knowledge of the
process prior to step k , and 𝑥𝑘 ∈ℜn to be our a
posteriori state estimate at step k given
measurement zk.
• We can then define a priori and a posteriori
estimate errors as
− −
𝑒𝑘 ≡ 𝑥𝑘 − 𝑥𝑘 &
𝑒𝑘 ≡ 𝑥𝑘 − 𝑥𝑘
• The a priori estimate error covariance is then
− − −𝑇
𝑃𝑘 = 𝐸 𝑒𝑘 𝑒𝑘
&
• the a posteriori estimate error covariance is
𝑇
𝑃𝑘 = 𝐸 𝑒𝑘 𝑒𝑘

• The posteriori state estimate 𝑥𝑘 is written as a


linear combination of an a priori estimate 𝑥𝑘−
and a weighted difference between an actual
measurement zk & a measurement prediction
H 𝑥𝑘− .
.
𝑥𝑘 = 𝑥𝑘− + 𝐾 𝑧𝑘 − 𝐻𝑥𝑘−

• The difference 𝑧𝑘 − 𝐻𝑥𝑘− is called the


measurement innovation, or the residual.

• The nxm matrix K is chosen to be the gain or


blending factor that minimizes the a posteriori
error covariance.

• Substituting 𝑥𝑘 in 𝑒𝑘 and 𝑒𝑘 in Pk , and


performing minimization, we get
𝑃𝑘− 𝐻𝑇 − 𝑇 − 𝑇 −1
𝐾𝑘 = = 𝑃𝑘 𝐻 𝐻𝑃𝑘 𝐻 + 𝑅
𝐻𝑃𝑘− 𝐻𝑇 + 𝑅
Kalman filter algorithm
• The Kalman filter estimates a process by using a form of
feedback control: the filter estimates the process state at
some time and then obtains feedback in the form of (noisy)
measurements.

• As such, the equations for the Kalman filter fall into two
groups: time update equations and measurement update
equations.

• The time update equations are responsible for projecting


forward (in time) the current state and error covariance
estimates to obtain the a priori estimates for the next time
step.
• The measurement update equations are responsible
for the feedback—i.e. for incorporating a new
measurement into the a priori estimate to obtain an
improved a posteriori estimate.

• The time update equations can also be thought of as


predictor equations, while the measurement update
equations can be thought of as corrector equations.

• The final estimation algorithm resembles that of a


predictor-corrector algorithm.
Discrete Kalman filter time update equations:

𝑥𝑘− = 𝐴𝑥𝑘−1 + 𝐵𝑢𝑘−1


Prediction
𝑃𝑘− = 𝐴𝑃𝑘−1 𝐴𝑇 + 𝑄

Discrete Kalman filter measurement update


equations:
𝐾𝑘 = 𝑃𝑘− 𝐻𝑇 𝐻𝑃𝑘− 𝐻𝑇 + 𝑅 −1

𝑥𝑘 = 𝑥𝑘− + 𝐾𝑘 𝑧𝑘 − 𝐻𝑥𝑘− Correction

𝑃𝑘 = 𝐼 − 𝐾𝑘 𝐻 𝑃𝑘−
A complete picture of the operation of the
Kalman filter
Implementation
• The image process is modelled as an auto
regressive(AR) process driven by a white
gaussian noise (Wn) with variance Q described
by

• Mathematically it can be written as


𝑦 𝑖, 𝑗
= 𝑎1 𝑦 𝑖, 𝑗 − 1 + 𝑎2 𝑦 𝑖 − 1, 𝑗
+ 𝑎3 𝑦 𝑖 − 1, 𝑗 − 1 + 𝑎4 𝑦(𝑖 − 1, 𝑗 + 1)
• The state space model for this system can be
written as
𝑋𝑛+1 = 𝐴𝑋𝑛 + 𝑉𝑛
𝑍𝑛 = 𝐻𝑋𝑛 + 𝑊𝑛

where

𝑎1 𝑎2 𝑎3 𝑎4 𝑦(𝑖, 𝑗 − 1)
1 0 0 0 𝑦(𝑖 − 1, 𝑗)
𝐴= & 𝑋𝑛 =
0 1 0 0 𝑦(𝑖 − 1, 𝑗 − 1)
0 0 1 0 𝑦(𝑖 − 1, 𝑗 + 1
Results
Original Images Measured Images Corrected Images
Extended Kalman Filter
• An extended Kalman filter is used if the
process to be estimated and (or) the
measurement relationship to the process is
non-linear.
• Here
𝑥𝑘 = 𝑓 𝑥𝑘−1 , 𝑢𝑘−1 , 𝑤𝑘−1

• with measurement z that is

𝑧𝑘 = ℎ 𝑥𝑘 , 𝑣𝑘
• Similar to the Kalman filter, the time and
measurement equations for EKF can be
written as below:
• EKF time update equations:

• EKF measurement update eqns:


A complete picture of the operation
of the extended Kalman filter
Results

Original Image Measured Image Corrected Image


COMPLEX KALMAN FILTERING
• In complex Kalman filtering, image model is
represented in complex form as real and imaginary
values represented as real and imaginary part of the
complex number.
Y= Real+(imag)i

• where, Y is complex image model.

• Complex valued Kalman filters have been used


extensively in a variety of applications, including
frequency estimation of time-varying signals, training
of neural networks etc.
Properties of Kalman filter
• Kalman filter is a time-varying filter as Kalman gain
changes with n.

• The filter is very powerful in several aspects: it


supports estimations of past, present, and even
future states, and it can do so even when the precise
nature of the modeled system is unknown.

• In the Kalman filter, prediction acts like the prior


information about the state at time n before we
observe the data at time n.
Refernces
• Natasha Devroye. Estimation: parts of Chapters 12-13,
Wiener and Kalman Filtering.

• Greg Welch and Gary Bishop. An Introduction to the


Kalman Filter, Monday, July 24, 2006.

• R. E. KALMAN. A New Approach to Linear Filtering and


Prediction Problems.

• http://www.cs.unc.edu/~welch/kalman/

You might also like