You are on page 1of 3

NATIONAL INSTITUTE OF TECHNOLOGY, HAMIRPUR

Department of Electronics and Communication Engineering


EC-743: Digital Image Processing and Pattern recognition (IInd Semester 2018-19)
Tutorial Sheet 1

Instructor: Dr. V. Prakash Singh Email: vpspeaks@gmail.com

1. (Point Processing/Intensity Transformation )


Decomposing an image into its bit planes is useful for analyzing the relative importance of each
bit in the image. The binary image for nth bit plane of an image is computed by processing
the input image with an intensity slicing transformation function that maps all intensities
which have nth bit as zero to some low level and all intensities which have nth bit as one to
some high level. Find and plot a set of intensity slicing transformation capable of producing
all individual bit planes for an 8 bit monochrome image.

2. ( Histogram based Processing)


Suppose a 3-bit image (L=8) of size 8 × 8 pixels has been read by a computer program and
stored in a matrix form as shown below.
 
1 1 1 1 1 1 1 1
 
 0 2 5 5 5 5 2 0 
 
 
 0 3 2 6 7 2 3 0 
 
 
 0 3 3 2 2 3 3 0 
I= 
 0 2 3 2 2 3 3 0 
 
 
 0 3 2 4 4 4 2 0 
 
 
 0 2 6 4 4 4 2 0 
 
1 1 1 1 1 1 1 1

(a) Draw the histogram of this image. Apply histogram equalization technique on this image
and show the equalized image, transformation function and histogram of the equalized
image.
(b) It is desired to transform the histogram of the input image so that it will have the values
specified in the table below. Plot the histogram of the histogram specified image.

zq 0 1 2 3 4 5 6 7
pz (zq ) 0.0 0.0 0.0 0.13 0.18 0.32 0.23 0.14

3. (Spatial Domain Processing)

(a) Consider the image shown in Fig. 1(a) which has equal number of gray (intensity value
0) and white (intensity value 252) pixels. An spatial filter mask h(x, y) that averages
the four closest neighbors of a point (x, y) but excludes the point itself from the average
is applied to this image. The output of averaging filter is further processed by a 3 × 3
median filter. Which intensity levels will possibly be present in the output image.
(Note that you do not need to compute the complete output image)

(b) Write the impulse response (mask) of 3 × 3 spatial filters for finding the components gx
and gy of gradient vector of an image f (x, y) at point (x, y). Using these filters, determine
the edge strength and direction of the edge at the marked point in the image shown in
Fig. 1(b). The gray pixel in the image have value 0 and the white pixels in the image
have value 1. (Note that the edge strength is the magnitude of the gradient vector and
the edge direction is perpendicular to the direction of the gradient vector).

(a) (b)

Figure 1: Figure for Problem 3

4. (Frequency Domain Processing)

(a) Consider an image f (x, y) of size M × N and a 3 × 3 spatial filter mask h(x, y) that is
used to implement the discrete 2-D Laplacian. Write an expression for spatial filtering
operation to get the Laplacian image g(x, y). Find an equivalent filter transfer function
in frequency domain, H(u, v), by expressing the Fourier transform of g(x, y) in terms of
Fourier transform of f (x, y). Will it be a lowpass filter or a highpass filter ? Explain.

(b) Suppose we wish to compute the Laplacian of the input image f (x, y) using frequency
domain filtering. Give a summary of all the steps required to accomplish this task.

5. (Spatial Domain Processing)

(a) Consider an image f (x, y) which is suffering from the following issues.

• The image does not have sufficient contrast and most areas in the image appear too
bright.
• The object boundaries in the image also appear blurred.
• There are random sparse black spots visible in the image.

Design an image processing system for enhancement of f (x, y) such that it addresses
all the aforementioned problems in the image. Explain each component of this image
enhancement system.

(b) What are isotropic filters ? Give a 3x3 spatial filter mask for any isotropic derivative
operator.

6. (Frequency Domain Processing)

(a) Consider an image f (x, y) of size M × N and a 3x3 spatial mask h(x, y that averages the
four closest neighbors of a point (x,y) including the point itself. Write the expression for
spatial filtering to get output image g(x, y). Find an equivalent filter transfer function
in frequency domain, H(u, v). Will it be a lowpass or a highpass filter ? Explain.

(b) What will be the effect of applying a filter with the following transfer function on an
image ?

H(u, v) = 0; where u = v = 0
H(u, v) = 1; elsewhere

You might also like