DIP Assignment 3

1. Explain homomorphic filter

2. Describe a image restoration model.
3. Discuss different noise models

4. Explain adaptive median filter.

5. Explain different order statistics filter to remove noise.
6. Explain uniform motion blur.
Uniform Motion Blur occurs when a camera or object moves at a constant speed in one direction
during exposure, causing the image to blur along the motion path. The blur is uniform in intensity,
creating streaks in the direction of motion, with no variation in the blur across the image.
Characteristics:
 Consistent direction and constant speed.
 Blurred streaks along the motion path.
 Common in photography with a slow shutter speed or in digital simulations.
Mathematical Representation:
The uniform motion blur is often modeled as a convolution with a motion blur kernel:

I(x,y)=1L∑t=0L−1I(x+t⋅cos⁡(θ),y+t⋅sin⁡(θ))I(x, y) = \frac{1}{L} \sum_{t=0}^{L-1} I(x + t \cdot \cos(\

theta), y + t \cdot \sin(\theta))
Where:
 I(x,y)I(x, y) is the image at coordinates (x,y)(x, y),
 LL is the length of the blur,
 θ\theta is the angle of motion,
 tt represents the distance of movement along the blur path.
Applications:
 In photography: Used to show motion or speed.
 In film/animation: Adds realism to fast-moving objects.
 In computer vision: Important for object tracking and recognition.
Handling:
 Restoration: Removing motion blur uses deconvolution techniques.
 Prevention: Use faster shutter speeds or image stabilization.
In essence, uniform motion blur creates streaked effects in the direction of motion, often used in
visual media to convey speed or action.

7. Explain RGB to HSI & HIS to RGB conversion

8. Explain properties of unitary transform
A unitary transform is a linear transformation that preserves the inner product of vectors, meaning
that it preserves the lengths (norms) and angles between vectors. Unitary transforms play an important
role in quantum mechanics, signal processing, and other fields of mathematics and engineering.
Here are the key properties of unitary transforms:
1. Preservation of Inner Product
 A transformation UU is unitary if it preserves the inner product between any two vectors. For
vectors x and y, this means:

⟨Ux,Uy⟩=⟨x,y⟩
 In other words, the angle between vectors and the magnitude (length) of vectors are preserved
under the transformation.
2. Length Preservation

 Unitary transforms preserve the norm (or length) of vectors. If ∥x∥ denotes the Euclidean
norm of a vector x, then for any vector x,

∥Ux∥=∥x∥
 This property ensures that unitary transformations do not stretch or shrink vectors.
3. Orthogonality
 A unitary matrix U satisfies the condition:
U†U=UU†=I
Where:
o U† is the conjugate transpose (Hermitian transpose) of U,

o II is the identity matrix.

 This means that the columns (and rows) of a unitary matrix are orthonormal (they are
orthogonal to each other and have unit length).
4. Invertibility
 A unitary matrix is always invertible, and its inverse is its conjugate transpose:
U-1=U†
 Since U†U= I, multiplying a unitary matrix by its conjugate transpose results in the identity
matrix, confirming that U-1=U†.
5. Eigenvalues on the Unit Circle
 The eigenvalues of a unitary matrix all lie on the unit circle in the complex plane. That is, for
any eigenvalue λ\lambda of a unitary matrix U,

∣λ∣=1
 This implies that the eigenvalues are of the form eiθ where θ is a real number.
6. Determinant
 The determinant of a unitary matrix has an absolute value of 1:

∣det(U)∣=1
 This follows from the fact that unitary matrices preserve the norm and volume, and their
determinant represents a scaling factor (which is 1 for unitary matrices).
7. Preservation of Orthogonality

 If UU is a unitary matrix, and x1and x2are orthogonal vectors (i.e., ⟨x1,x2⟩=0 ,then their
transformed counterparts Ux1 and Ux2 will also be orthogonal:

⟨Ux1,Ux2⟩=0
8. Unitarity in Quantum Mechanics
 In quantum mechanics, unitary transforms represent quantum operations (such as quantum
gates in quantum computing) that preserve the total probability amplitude of quantum states.
The preservation of inner product implies that probabilities (squared magnitudes of quantum
states) remain unchanged under these transformations.
9. Matrix Representation
 A unitary matrix UU is typically represented as a square matrix with complex entries. For
example, a 2x2 unitary matrix can be written as:

where the complex entries satisfy the condition U†U= I.

10. Commutativity with Hermitian Matrices
 A unitary matrix U commutes with a Hermitian matrix H (i.e., UHU†=H) if and only if U is
diagonalizable by H.
9. Explain the properties of 2D DFT.
10. Construct haar transform for N=3.
11. Explain wiener filter.
12. List out the properties of Hadamard transform
The Hadamard Transform is a mathematical operation often used in signal processing, image
processing, and quantum computing. It is based on the Hadamard matrix, which is a square matrix
whose entries are either +1 or -1. The Hadamard transform has several important properties:
1. Linearity
 The Hadamard transform is linear, meaning it satisfies the principles of superposition and
scaling. If a and b are vectors and cc is a scalar, then:
H(a+b)=H(a)+H(b)
H(ca)=cH(a)
2. Orthogonality
 The rows and columns of the Hadamard matrix are orthogonal to each other. Specifically:
HT H=HHT =nI
Where HTH^T is the transpose of the Hadamard matrix, n is the size of the matrix, and I is
the identity matrix. This orthogonality means that the Hadamard transform preserves inner
products.
3. Involutory (Self-Inverse)
 The Hadamard transform is involutory, meaning applying the transform twice returns the
original vector. Mathematically:
H(H(x))=x
 This property is similar to the Fourier transform and is crucial for various applications like
error correction and image processing.
4. Efficient Computation
 The Hadamard transform can be computed efficiently using a divide-and-conquer algorithm,
similar to the fast Fourier transform (FFT). The computational complexity is O(n log n),
making it very efficient for large data sets.
5. Energy Preservation (Parseval's Theorem)
 The Hadamard transform preserves the energy of the signal. According to Parseval's
Theorem, the sum of the squared elements of a vector is the same before and after applying
the transform:

This ensures that the transform does not distort the magnitude of the signal.
6. Binary Operation
 The Hadamard transform operates on binary values (+1+1 and −1-1), which is particularly
useful in applications like error correction codes, where the coefficients are taken from these
values.
7. Symmetry
 The Hadamard matrix exhibits a recursive structure, meaning it can be constructed using
smaller Hadamard matrices. Specifically, the Hadamard matrix H 2n can be built from Hn as:

 This structure provides symmetry and scalability, allowing efficient construction and
computation for larger matrices.
8. Application in Quantum Computing
 The Hadamard transform plays a key role in quantum computing, where it is used to create
superposition states. It is used to map basis states in quantum algorithms, such as the
Deutsch-Jozsa algorithm and Grover's search algorithm.
9. Decoding in Error-Correction Codes
 The Hadamard transform is widely used in error-correction codes, such as Hadamard
codes, which help detect and correct errors in data transmission systems.
10. Determinant
 The determinant of a Hadamard matrix is either +1+1 or −1-1, depending on the size of the
matrix. For example:

13. Explain the concept of optimum notch filtering.

Optimum Notch Filtering is a signal processing technique used to remove or attenuate a specific
frequency (or narrowband interference) while preserving the rest of the signal. It operates in the
frequency domain and creates a "notch" or deep dip at the target frequency, leaving other frequencies
unaffected.
Key Concepts:
1. Frequency Domain Filtering: The signal is decomposed into its frequency components, and
a notch filter selectively reduces the amplitude of the target frequency.
2. Notch Filter Design: The filter has a narrow bandwidth and deep attenuation at the center
frequency. The key parameters include:
o Center frequency (f₀): The frequency to be attenuated.

o Bandwidth (Δf): The width of the notch.

o Depth of attenuation: How much the target frequency is reduced.

3. Optimum Design: The goal is to achieve the best attenuation at the target frequency with
minimal impact on the rest of the signal. The filter should minimize distortion and not affect
neighboring frequencies.
4. Applications:
 o Removing power line interference (50/60 Hz) in medical and audio systems.

o Filtering narrowband interference in communication systems.

5. Mathematical Representation: A simple transfer function for a notch filter is:

Where f0 is the center frequency, and Δf is the bandwidth.

Practical Considerations:
 Trade-off: Narrower notches reduce signal distortion but may not attenuate the target
frequency enough. Wider notches provide better attenuation but may distort the signal.
 Filter order: Higher order filters provide sharper attenuation but increase complexity.

14. Discuss RGB and CMY color cube.

RGB and CMY Color Cubes
The RGB (Red, Green, Blue) and CMY (Cyan, Magenta, Yellow) color models are fundamental in
understanding how colors are represented in digital systems, and both can be represented using a 3D
color cube. These models are based on the primary colors used in light and pigments, respectively,
and are commonly used in various display technologies and printing systems.

RGB Color Cube

1. Primary Colors:
o Red (R), Green (G), and Blue (B) are the primary colors in the additive color
model, which is used in displays (monitors, TVs, etc.).
o In the RGB model, colors are created by combining different intensities of these three
colors of light.
 2. Color Cube Representation:
o The RGB color cube is a 3D representation of all possible colors that can be created
by varying the intensities of the red, green, and blue components.
o Each axis of the cube represents one of the primary colors (R, G, and B).

o The cube typically has values from 0 to 255 for each axis in 8-bit color depth, with 0
meaning no intensity and 255 being full intensity.
o Corners of the cube correspond to pure colors (e.g., (255, 0, 0) for red, (0, 255, 0) for
green, and (0, 0, 255) for blue).
o Center: (127, 127, 127) corresponds to a gray color (balanced intensity of all three
primary colors).
o White: (255, 255, 255) is the combination of maximum intensity of all three colors.

3. Color Mixing:
o In the RGB model, combining two primary colors creates secondary colors (e.g., red
and green create yellow, green and blue create cyan).
o When all three components are at full intensity, the result is white. If all three are at
zero, the result is black.

CMY Color Cube

1. Primary Colors:
o The CMY model uses Cyan (C), Magenta (M), and Yellow (Y) as its primary
colors, which are the complementary colors to red, green, and blue.
o The CMY model is used primarily in subtractive color mixing, like in printing,
where colors are created by subtracting light.
2. Color Cube Representation:
o The CMY color cube is also a 3D space, but instead of additive mixing (as in RGB),
it represents subtractive mixing.
o Each axis of the cube corresponds to the cyan, magenta, and yellow components.

o The color space is inverted compared to the RGB model: higher values indicate more
color saturation (with 0 representing no color, and 255 being maximum saturation).
o Pure Cyan corresponds to (255, 0, 0), Pure Magenta corresponds to (0, 255, 0), and
Pure Yellow corresponds to (0, 0, 255) in the CMY space.
o White is (0, 0, 0), and Black is (255, 255, 255) in the CMY cube (the reverse of the
RGB cube).
3. Color Mixing:
o In the CMY model, mixing the primary colors results in the creation of secondary
colors. For example, cyan and magenta mix to form blue, magenta and yellow form
red, and yellow and cyan form green.
o When all three components are mixed at full intensity, the result is black (in theory,
but practical systems use a fourth component, Key (K), in the CMYK model to
improve the depth of black).