Sei sulla pagina 1di 25

MUHAMMED MUSTAFA TARHAN & OUZHAN DEMRC

Image Processing
The aim of this lab acquiring experiments about :
Image Pre-processing.

Morphological operations.
Filter techniques.

With using matlab

Section 1 Preparing images


In this section we prepared images for the processing. Preperations include: neighborhood operation pre-processing for enhancement. Edge detection with our algorith. with using matlap.

Neighborhood operation
In this operation each output pixel is computed from a

set of neighboring input pixels.

Result of neighborhood
In this lab, we implamented [3 3]window silinding

process and block process using matlab. As a result central pixel have the mean value.

Result of neighborhood

Pre-processing
For the processing we applied: Histogram analysis Avarage filtering for the noise reduction median filter to reduction of blurring for enchancement.

Original Image and Histogram


In the left histogram, you can see intensity levels are getting high in the white place. When it is on the coin there is less intensity.
Noise Type In the Image : Salt && Pepper

Removed Noise and Histogram

As you can see from the removed noise figure, we have minimize the noise by using fspecial matlab function and to make it more clear we used median filtering function to blur it.

SECTION 2 MORPHOLOGICAL OPERATIONS


In this section we applied morphological implamentations such as: structure element Labelling Edge thicknissing.

Three big circles


In this task we convert intensity image to gray image.
we applied labelling to segmented image. we implamented labelling to the the circles.

Also we applied some morphological operations for

tickness egdes of the images. We got three big objects by using bwareaopen.

RESULT

Music notes
In this task we use some morphological processes for

removing staff lines and for getting staff notes such as: we took complament of original image. We cropped staff lines. We increased thicness of the notes as horizontally and vertically. We converted image to binary image form We applied edge detection to notes. We filled the edges of nodes. Since the edges are not touching each other, we could not get the whole result we only remove the lines and some other small objects.

Operations to connect seperate edges to each other

The Image Only removed Staff lines

Section 3 Filter Techniques


In this section we applied filters such as.
Spatial domain filter:
Masking. Statistical filter, median filter. Unsharp masking.

Frequency domain filter:


Lowpass filter. Highpass filter.

High-Frequency emphasis filter.

Task 1 Reduction of Blurring


We use high-pass filter to decreas blurring and

increase sharpness by attenuation the low frequency and leaving high frequencies. Before filtering the images, we converted image from spatial domain to frequency domain with fast fourier transform.

Results...

Removing DC Component
The value of the transform at the origin of the

frequency domain is called for ex (F[0 0]) DC component of the faurier transform where DC signifies direct current (current of zero frequency). To remove DC component on the image, we shift the first [0 0] to the center then, we equalize that value to the zero to compress DC. After getting the result, we shifted our origin back to the [0 0]

Results

Task 3 Paper Surface


In this task we used fourier transform to find the

frequency componanet which responsable of the texture on surface . We implement low pass filter for smooting .

Applying Low pass filter


Basic steps for low pass filtering
1) obtaining padding size parameters using paddedsize 2) 3) 4) 5) 6)

functions. (It is given in the book) Obtaining the fourier transform with padding. We generated filter function Multiplication of tranfer function by the filter Obtaining the real part of the invers FFT of multiplication results Cropping the top , left rectangle to the original size.

Results

Result of fft

Result of the smoothen paper surface

Potrebbero piacerti anche