Q1. Write a R program for Point-to-point transformation.
This laboratory
experiment provides for thresholding an image and the evaluation of its histogram.
Histogram equalization. This experiment illustrates the relationship among the
intensities (gray levels) of an image and its histogram.
# Program:
# Load necessary packages
library(imager)
library(ggplot2)
# Load an image
img <- load.image("path/to/your/image.jpg")
# Convert the image to grayscale
gray_img <- grayscale(img)
# Apply histogram equalization
equalized_img <- equalize(gray_img)
# Compute histograms
hist_original <- hist(as.vector(as.matrix(gray_img)), breaks = seq(0, 1, by = 1/256),
plot = FALSE)
hist_equalized <- hist(as.vector(as.matrix(equalized_img)), breaks = seq(0, 1, by =
1/256), plot = FALSE)
# Plot the histograms
ggplot() +
geom_line(aes(x = hist_original$mids, y = hist_original$counts), color = "blue") +
labs(title = "Original Image Histogram") +
theme_minimal()
ggplot() +
geom_line(aes(x = hist_equalized$mids, y = hist_equalized$counts), color = "blue") +
labs(title = "Equalized Image Histogram") +
theme_minimal()
# Display the original and equalized images
par(mfrow = c(1, 2))
plot(gray_img, main = "Original Image", col = grey.colors(256))
plot(equalized_img, main = "Equalized Image", col = grey.colors(256))
Q2. Write a R program for Geometric transformations. This experiment shows
image rotation, scaling, and translation. Two-dimensional Fourier transform.
# Program:
1 Image Rotation, Scaling and Translation:
library(imager)
library(ggplot2)
# Load an image
image <- load.image('path/to/your/image.jpg')
# Rotation
rotation_matrix <- matrix(c(cos(pi/4), -sin(pi/4), sin(pi/4), cos(pi/4)), nrow = 2)
rotated_image <- imager::rotate(image, matrix = rotation_matrix)
# Scaling
scaling_factor <- 1.5
scaled_image <- resize(image, scale = scaling_factor)
# Translation
translation_matrix <- matrix(c(1, 0, 50, 0, 1, 30), nrow = 2)
translated_image <- translate(image, matrix = translation_matrix)
# Display the original and transformed images
par(mfrow=c(2, 2))
plot(image, main = 'Original Image')
plot(rotated_image, main = 'Rotated Image')
plot(scaled_image, main = 'Scaled Image')
plot(translated_image, main = 'Translated Image')
2 Two-dimensional Fourier Trasnsform:
# Load necessary packages
library(imager)
library(ggplot2)
# Load an image
image <- load.image('path/to/your/image.jpg')
# Convert the image to grayscale
gray_image <- grayscale(image)
# Compute the 2D Fourier Transform
fourier_transform <- fft(as.cimg(gray_image))
# Compute the magnitude spectrum
magnitude_spectrum <- log(abs(fourier_transform) + 1)
# Display the original image and its magnitude spectrum
par(mfrow=c(1, 2))
plot(gray_image, main = 'Original Image', col = grey.colors(256))
plot(magnitude_spectrum, main = 'Magnitude Spectrum', col = grey.colors(256))
Q3. Write a R program for Linear filtering using convolution. Highly selective filters.
# Program:
1 Linear filtering using convulation:
library(imager)
library(ggplot2)
# Load an image
image <- load.image('path/to/your/image.jpg')
# Convert the image to grayscale
gray_image <- grayscale(image)
# Define a Gaussian filter (Example: Highly selective filter for blurring)
gaussian_filter <- matrix(c(1, 4, 6, 4, 1,
4, 16, 24, 16, 4,
6, 24, 36, 24, 6,
4, 16, 24, 16, 4,
1, 4, 6, 4, 1), nrow = 5, byrow = TRUE) / 256
# Apply convolution with the Gaussian filter
filtered_image <- convolve(gray_image, gaussian_filter, boundary = "wrap", normalize
= TRUE)
# Display the original and filtered images
par(mfrow=c(1, 2))
plot(gray_image, main = 'Original Image', col = grey.colors(256))
plot(filtered_image, main = 'Filtered Image', col = grey.colors(256))
2 Highly selective filter example: Edge Detection (Sobel filter)
# Load necessary packages
library(imager)
library(ggplot2)
# Load an image
image <- load.image('path/to/your/image.jpg')
# Convert the image to grayscale
gray_image <- grayscale(image)
# Apply Sobel filter for edge detection
sobel_x <- imager::sobel(gray_image, direction = "x")
sobel_y <- imager::sobel(gray_image, direction = "y")
magnitude <- sqrt(sobel_x^2 + sobel_y^2)
# Display the original and edge-detected images
par(mfrow=c(1, 2))
plot(gray_image, main = 'Original Image', col = grey.colors(256))
plot(magnitude, main = 'Edge-Detected Image', col = grey.colors(256))
Q4. Write a R program for Ideal filters in the frequency domain. Non Linear filtering
using convolutional masks. Edge detection. This experiment enables students to
understand the concept of edge detectors and their operation in noisy images.
# Program:
1 Ideal Filters in Frequency domain
library(imager)
library(ggplot2)
# Load an image
image <- load.image('path/to/your/image.jpg')
# Convert the image to grayscale
gray_image <- grayscale(image)
# Compute the 2D Fourier Transform
fourier_transform <- fft(as.cimg(gray_image))
frequencies <- fftshift(fourier_transform)
# Create an ideal high-pass filter
rows <- nrow(gray_image)
cols <- ncol(gray_image)
center_row <- rows / 2
center_col <- cols / 2
radius <- 30 # Adjust the radius based on your requirements
mask <- matrix(1, nrow = rows, ncol = cols)
mask <- cv2.circle(mask, c(center_col, center_row), radius, 0, -1)
# Apply the ideal high-pass filter
frequencies_high_pass <- frequencies * mask
filtered_image_high_pass <- Re(fft(as.cimg(frequencies_high_pass), inverse = TRUE))
# Display the original and filtered images
par(mfrow=c(1, 2))
plot(gray_image, main = 'Original Image', col = grey.colors(256))
plot(filtered_image_high_pass, main = 'Filtered Image (High Pass)', col =
grey.colors(256))
2 Non-linear Filtering for Edge Detection (Using Laplacian Mask)
# Load necessary packages
library(imager)
library(ggplot2)
# Load an image
image <- load.image('path/to/your/image.jpg')
# Convert the image to grayscale
gray_image <- grayscale(image)
# Apply a Laplacian mask for edge detection
laplacian_mask <- matrix(c(0, 1, 0,
1, -4, 1,
0, 1, 0), nrow = 3, byrow = TRUE)
filtered_image_laplacian <- convolve(gray_image, laplacian_mask, boundary =
"wrap", normalize = TRUE)
# Display the original and filtered images
par(mfrow=c(1, 2))
plot(gray_image, main = 'Original Image', col = grey.colors(256))
plot(filtered_image_laplacian, main = 'Edge-Detected Image (Laplacian)', col =
grey.colors(256))
