EN3160 - Intensity Transformations and Neighborhood Filtering

This repository focuses on Intensity Transformations and Neighborhood Filtering where it involves implementing various image processing techniques on a set of images. This was done as part of the module Image Processing and Machine Vision- EN3160.

Table of Contents

1. Dependencies
2. Tasks
   • Intensity Transformation
   • Accentuate Brain Matter
   • Gamma Correction
   • Vibrance Enhancement
   • Histogram Equalization
   • Foreground Histogram Equalization
   • Sobel Filtering
   • Image Zooming
   • Image Segmentation and Enhancement
3. Results

Dependencies

This project requires the following Python libraries:

• OpenCV
• NumPy
• Matplotlib

You can install them using pip:

pip install opencv-python numpy matplotlib

Tasks

Intensity Transformation

• Task: Implement the piecewise linear intensity transformation on the provided image of Emma Watson.
• Files: q1.py, a1images/emma.jpg

Accentuate Brain Matter

• Task: Apply a similar piecewise linear transformation to accentuate the white and gray matter in a brain proton density image. The transformations are shown as plots.
• Files: q2.py, q2_histogram.py, a1images/brain_proton_density_slice.jpg

Gamma Correction

• Task: Apply gamma correction to the L plane of an image in the Lab* color space. The gamma value is adjusted, and histograms of the original and corrected images are shown.
• Files: q3.py, a1images/highlights_and_shadows.jpg

Vibrance Enhancement

• Task: Increase the vibrance of a photograph by applying a given intensity transformation to the saturation plane. The process involves splitting the image into H, S, V planes, applying the transformation, and recombining them.
• Files: q4.py, a1images/spider.png

Histogram Equalization

• Task: Write a custom function to perform histogram equalization on an image. The histograms before and after equalization are displayed.
• Files: q5.py, a1images/shells.tif

Foreground Histogram Equalization

• Task: Apply histogram equalization only to the foreground of an image. This involves creating a mask, separating the foreground and background, equalizing the foreground, and then merging them back.
• Files: q6.py, q6_optional.py, a1images/jeniffer.jpg

Sobel Filtering

• Task: Compute the gradient of an image using the Sobel operator in three different ways: using the filter2D function, writing custom code, and using the separability property of the Sobel kernel.
• Files: q7.py, q7_2.py, q7_3.py, a1images/einstein.jpg

Image Zooming

• Task: Write a program to zoom images by a given factor using both nearest-neighbor and bilinear interpolation methods. The results are tested by comparing scaled-up small images with their original larger versions.
• Files: q8.py, a1images/a1q5images/im01small.png, a1images/a1q5images/im01.png

Image Segmentation and Enhancement

• Task: Use the GrabCut algorithm to segment a flower image. Then, produce an enhanced image with a substantially blurred background.
• Files: q9.py, a1images/daisy.jpg

Results

All the output images, plots, and histograms can be found in the /Results directory.