Digital Image Processing

Repo for assignments done in ELL715: Digital Image Processing @ IIT Delhi

Description

In each assignment folder, there's the problem statement, the code and the final report submitted.

The main tasks performed in the assignments are:

1. Image Transformations:
   • Polar transformation
   • Shear transformation
   • Geometric rotation, scaling and translation
2. Image Histogram and Equalization:
   • Computed histogram of an RGB image
   • Improved contrast of a grayscale image using:
     • Contrast Stretching
     • Gamma Correction
     • Histogram Equalization
     • Unsharp masking
3. Image Filtering and Scaling:
   • Analyzed the effect of an IIR filter and unsharp masking (shaprening filter)
   • Played with fourier transform and inverse fourier transform
   • Performed Discrete Cosine Transform on images
4. Image Compression:
   • Implemented bilinear, bicubic, DCT and FFT compression
   • Implemented the Seam carving by removing areas with low energy
   • Impelemnted compression using 2-D Wavelet decomposition
5. Wavelet Analysis:
   • Extracted retinal blood vessels from fundus images using:
     • Wavelets
     • Sobel
     • LoG
6. Image Morphing:
   • Separated occluded objects using morphology and thresholding
   • Counted circular objects in an image
   • Detected black note heads
7. Vessel Segmentation using Morphology:
   • Removed the bright, circular shaped Optic Disc
   • Extracted the vessel map
   • Estimated the diameter of thinnest & thickest vessel
   • Removed small isolated patches and smoothened the vessel map
8. Image Segmentation:
   • Implemented segmentation using Intensity based thresholding
   • Implemented segmentation using Region growing
   • Implemented segmentation using Image Matting
9. Optic Disc Segmentation Using Active Contour:
   • Imeplemented algorithm to find the center of the optic disc
   • Developed an active contour algorithm to fit the initial contour at the edges of the disc
   • Removed small isolated patches and smoothened the vessel map
10. Retinal Vessel Segmentation Using Neural Networks:
    • Extracted features for vessel map segmentation
    • Followed leave-one-out strategy to train the neural network
    • Evaluated the performance of algorithm in terms of:
      • Accuracy
      • Sensitivity
      • Specificity
      • Area Under Receiver Operating Characteristics Curve (AUC under ROC)

Credits

Most of the work here was done in collaboration with @naman1122000