ADAPTIVE BILATERAL FILTER FOR SHARPNESS ENHANCEMENT AND NOISE REMOVAL

By,

AIM:

The main aim of our project is to implement the adaptive bilateral filter (ABF) algorithm for sharpness enhancement and noise removal.

NECESSITYThe goal of image restoration is to improve a degrade image in some predefined sense. Schematically this process can be visualized as

where f is the original image, g is a degraded/noisy version of the original image and f ̃ is a restored version.

Scope

The scope of this project is to deal with images that are appropriate for digital

photography. We do not consider images that are severely

degraded.

Importance of technique The ABF sharpens an image by increasing the slope of the

edges without producing overshoot or undershoot. It is an approach to

sharpness enhancement that is fundamentally different from the unsharp

mask (USM). This new approach to slope restoration also

differs significantly from previous slope restoration algorithms in that

the ABF does not involve detection of edges or their

orientation, or extraction of edge profiles.

• Image Enhancement: – A process which aims

to improve bad images so they will

“look” better. No quantitative measures Subjective Remove effects of sensing environment

Image Restoration: – A process which aims

to invert known degradation operations

applied to images. Mathematical, model dependent quantitative measures• Objective

Enhancement v.s. Restoration

Image restration :

Image restoration methods are used to improve the appearance of an image by application of a restoration process. It is use mathematical model for image degradation.

Eg:-1.Geomatric distortion caused by imperfect lenses.2.Superimposed interference patterns caused by

mechanical system.3.Noise from electronics sources.

The Goal of Image Restoration

The Degrade Image:- f(x,y) F=Original Image G= Degrade Image F=restoed The ABF are sharpens an Image by increasing the also slope

of the edges without undershoot (or) overshoot. Degradation models:- Motion Blur: long exposure uniform 2D Blur Out-Of-Focus Blur

Degradation Models

Degradation ModelsImage degradation can occur for many reasons, some typical degradation models are

Motion Blur: due to camera panning or subject moving quickly.

Atmospheric Blur: long exposure

Uniform 2D Blur

Out-of-Focus Blur

Ideally the value of a pixel should be the light intensity at a

infinitesimal point in the imaged scene.

Each sensor in a CCD array integrates the light intensity in a small

area surrounding a point with possibly non-equal weighting

The point spread function (PSF) is the image captured when there

is only one single point with high intensity in the scene.

This PSF is the degradation filter.

Typically h(x,y) due to sensor PSF is low-pass and is often

approximated by a Gaussian filter – h(x,y)= e^-k (x^2+y^2

Degradation Models

Cause of blurring :

The blerring (or) degradation of image can be caused by many factors.

1.The image capture process :by the camera (or) when long supose times are used.

2.out –of –focus optics :used as a mwide-angle lens.which are to reduced the number of photos captured.

Noise removal using spatial fillters :it is used to remove the various type of noise in digital image.it operated the small neighbouroods,3*3 to11*11.

Typical PSF Blurness

Degradation Due to Motion Blur

The sensor integrates the intensity value of a point over a certain exposure time T

When there are moving objects in the imaged scene at high speed (relative to exposure time), we see motion blur.

To reduce the motion blur, one can reduce the exposure time

Often due to camera panning or fast object motion. Linear along a specific direction.

Degradation Due to Motion Blur

original image

20 40 60

10

20

30

40

50

60

blurred image

20 40 60

10

20

30

40

50

60

Noise Models

Most noise models assume the noise is some known probability density function. The density function is chosen based on the underlining physics.

Gaussian: poor illumination.

Rayleigh: range image

Salt andPepper: faulty switch during imaging

Gamma or Exp: laser imaging

There are two types of filters:- 1.Analog: using analog electronics circuits such as

resistor, capacitor,opamp. These are using the noise reduction.

2.digital filters: these are using digital processor to

perform numerical calculation on sampled value of the signals.

Order filter :

Order filter operates on small sub image the most useful of the order filters in the median filter.

The median filter select the middle pixel value from the order set.

These disadvantages eliminate adaptive median filter. Order of the filter are two

1.Maximum filter :It is used to the selected the longest value in the window of the pixel.

2.Minimum filter :It is used to the selected the smallest values of the window of the pixel.

BILATERAL FILTER

Bilateral filtering smoothes images while preserving edges, by means of a nonlinear combination of nearby image values.

Domain filteringc(ξ,x) measures the geometric closeness between the neighborhood center x and a nearby point ξ

Range filterings( f(ξ), f (x)) measures the photometric

similarity between the pixel at the center point x and that of a nearby point ξ

Piecewise-linear bilateral filtering

Suppose there is n pixels in a image, a bilateral filtering might require O(n2) time – VERY SLOW.

A filtering method with fixed mask like mean filtering is a convolution by mask and image.

The mask of bilateral filtering varies with its center pixel.

1 1 1

1 1 1

1 1 1

Piecewise-linear bilateral filtering (cont.)

Divide a image into M level by intensity; each level has a reference intensity fj . So we have a new filtering funtion:

This new function has a fixed mask within each level.

The final output h(x) is a linear interpolation between the output hj(x) of the two closest values fj .

How to decide σr by image ? Noise causes smaller

difference between pixels, and edges cause lager difference.

Maybe we can evaluate the difference within image to find a better σr.

How to Compute Every Pixel

f1 f2 f3

f4 f5 f6

f7 f8 f9

×w1 w2 w3

w4 w5 w6

w7 w8 w9

∑f(i)*w(i)

Origin Image Weight Output Image

Normalisation Factor

* =

Bilateral Filter For image I at coordinate I(m,n):

geometric closeness between (m,n) and (m0,n0)

photometric similarity between I(m,n) and I(m0,n0)

Bilateral FilterImpulse

ResponseDegraded

ImageBilateral filter

output

BILATERAL FILTER

Advantages Fast Simple Intuitive parameter selection Edges preserving

Limitations Over smooth Can’t iterate many times

ADAPTIVE BILATERAL FILTER The ABF retains the general form of

a bilateral filter, but contains two important

modifications. First, an offset is introduced

to the range filter in the ABF. Second, both and the

width of the range filter in the ABF are locally adaptive

Adaptive Bilateral FilterImpulse

ResponseDegraded

ImageAdaptive

Bilateral filter output

Adaptive Bilateral Filter Advantages

Fast Simple Intuitive parameter selection Edges preserving Sharpens image Smoothes the noise

APPLICATIONS

Digital Photography Astronomical image restoration

Algorithm for Bilateral Filter

Take image as input and also range filter and domain filter values

Calculate Gaussian weights Calculate a smaller window Calculate normalization factor for the window Calculate the impulse response of the

window Repeat the process from third step till

thesize of whole image Send the image as output

Algorithm using ROI Read an Image Convert the RGB image to Gray Calculate ROI Form a degraded image by adding

PSF blur and random noise Apply Bilateral Filter Apply Adaptive Bilateral Filter Display the Images

Algorithm for whole image

Read an Image Form the Degraded Image Apply Bilateral Filter Apply Adaptive Bilateral Image Display the Images

Algorithm for Adaptive Bilateral Filter Take image as input and also range filter and

domain filter values and gamma values Calculate Gaussian weights Calculate a smaller window Calculate normalization factor for the window Calculate the impulse response of the

window Repeat the process from third step till the

size of whole image Send the image as output

Creating the ROI (Region Of Internet)

The ROI are using the analysis on past of an Image. The ROI are actually using the standards' counters or) free counters.

an image most is 8-bit image. If a pixel in the mask

is non zero the corresponding pixel in the image to process is processed.

If the pixel in the image mask value is ‘0’ the

corresponding pixel in the image to process in left unchange. The default value of the offset is (0,0).

ROI Containg The two Elements:-

1. Bounding rectangle for an ROI: 2.Regions List: these array are containing counter

identifier. Where as 0An exterior counter 1An interior counter.

Introduction to digital filters:- In a signal processing the function of filter is to

remove unwanted past of the signal such as random noise.

FLOW CHART

Degraded image

Knowledge of image creation process

Input imaged(r,c)

Develop Degradedmodel

Develop inverseDegradedprocess

output imageI(r,c)

Apply inverse degradation

process

Linear filters:- you can use linear filtering to remove certain types of noise such as Gauss ion filters.

Non-Linear Filters:- It is extended linear filter. It is using the to find the rank and order.

Introduction to MATLAB:- MATLAB is high-performance language for technical computing. It

integrated computation visualization and programming. The program in written in non-interactive language such as a C (or)

FORTRAN.

The MATLAB System :

It is contain two 5 main parts. 1.Develpment Environment

2.The MATLAB mathematical function

3.The MATLAB language

4.Graphics

5.The MATLAB application program interface(API)

Requirements

Computer 40 GB Hard Disk Pentium Processor 2.6 or 3.0 G Hz SpeedSoftware:MATLAB 7.2 or 7.4 or 7.6 version

DEGRADED RESULTSDegraded ImageOriginal Image

HISTROGRAM

0

100

200

300

400

500

600

700

800

histogram Degraded Image

0 50 100 150 200 250

BILATERAL RESULTSOriginal ImageBilateral filter image

0

100

200

300

400

500

600

700

800

histogram bilteral filtered Image

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

ADAPTIVE BILATERAL RESULTS

Adaptive bilateral FilterAdaptive bilateral Filter

0

100

200

300

400

500

600

700

800

histogram adaptive bilteral filtered Image

0 50 100 150 200 250

ROI RESULTS

original iumage

FUTURE SCOPE

For future development of the ABF, we would suggest that the following

issues be addressed.

First, the ABF tends to resize the image, due to its fundamental

mechanism of sharpening an image by pulling up or pushing down pixels

along the edge slope.

Second, the ABF does not perform as well at corners as it does on

lines and spatially slow-varying curves, since the ABF is primarily based

on transforming the histogram of the local data, which cannot effectively

represent 2-D structures.

Finally, in the current design of the ABF, a fixed domain Gaussian

filter is used.