DIP Realized by IDL Author: Ying Li Course: computer for imaging science.

DIP Realized by IDL

Author: Ying Li

Course: computer for imaging science

Program Overview

My project has 5 modules: 1. Zooming module 2. Filter module 3. Fourier transform module 4. Histogram module 5. Motion blur and restoration module

Zooming Module Interface

Filter module Interface

Fourier transform module

Histogram module interface

Motion blur and restoration

Interface Architecture

Selection Base

Menu bar

Zoomingbase

Displaywindow

Zooming module

Zooming Module

Color table Keep track of the button status Restrict the rectangle from going

outside of the image Erase old rectangles

Color Table

I want to display a gray level image with a red region of interest displayed on it. So I need a color table of my own.

I set a flag variable to keep track of the mouse buttons’ status. So user can only drag the red rectangle with the mouse button pressed down.

The program calculated carefully to prevent the red rectangle from going outside of the image.

Erase old rectangle

In order for the red rectangle to go with the mouse, the program must erase the old rectangle and draw a new rectangle at the new position.

To do this I use a hidden draw widget to display the image at exactly the same position, and erase old rectangle by copy data from the hidden window.

Filter Module

In this module I realized four kind of filters:

Ideal low pass filter Ideal high pass filter Ideal band pass filter Butterworth low pass filter

Ideal low pass filter

Ideal high pass filter

Butterworth filter

Butterworth Filter:

We know because of the the sharp edge of the ideal filters, there will be some oscillation on the output signal of ideal filters.

This is the output of a STEP function go through an ideal low pass filter

So, we want a kind of filter whose edges go down slowly. Butterworth filter was introduced.

This is the equation of a 1-D Butterworth filter:

Here N is the order of the Butterworth filter and c is the frequency cutoff

Nc

B2)/(1

1|)(|

N=2 N=6

Restore this degraded image:

Go through a ideal low pass filter

Go through a Butterworth filter

Fourier Transform Module

Histogram module

Histogram module

Motion Blur Restoration

Using a Inverse Filter to deconvolve the point spread function

Using convolve method to get ride of the blur coursed by the motion of the detector or the object

Inverse Filter

Before image restoration can be accomplished, the PSF of the blurring function(that is the system transfer function of the degrading system) must be known. Actually most system that course the degrading of images are linear shift invariant system.

Degradation model:

Solve by inverse filter:

Here if the noise is very small and can be neglected. Then we can restore the image by a reverse filter:

),(),(),(),( NFHG

HGF /

PSF:

The point spread function is a line here, if the exposure time is small enough.

Result of inverse filter method:

Convolution Method:

We still have some other ways to restore a motion blurred image. The motion blur is coursed by the moving of the detector or the object within the exposure thim T. That is:

T

dttyytxxgyxf0 00 ))(),(().(

Convolution method:

iterate the procedure we can get the follow equation:

)()(')(')(

)()(')(

)()()('

xxfaxfaxg

xxfxg

axgxgxf

m

l

xlaxfmaxg0

)()(')(

Convolution method:

From that we can see that the result is the convolution of the derivation of the degraded image with a comb function

Result of convolution method

Result of convolution method

Result of convolution method

Result of convolution method

Conclusion:

In this project I used such widgets: labels, texts, draws,bases,drop lists, radio buttons, slider bars, menus, module dialogue form.

I realized such functions: Region of interest, ideal low pass filter, ideal high pass filter, ideal band pass filter, butterworth filter, with different parameters, fourier transform, image histogram, histogram equalization, image blur, a inverse filter, convolution method to restore motion blur, a module dialogue form