You are on page 1of 16

Question 1(a)

Full Scale Contrast Stretch for different images and respective Histograms after full scale contrast
stretching
Question 1(b): Histogram Equalization
Histogram equalized images and their respective histograms after histogram equalization

Fig1 Fig2

Fig3 Fig4

Fig 5
Question1(c) Non-Linear contrast stretch:
Histogram equalized Image and respective Histogram (after equalization)
Fig1 : IISc Main building Image

Fig2: LowLight2 image


Fig3: LowLight_3 Image

Fig4 : Hazy Image


Fig5 : StoneFace Image
Question1(d): CLAHE
For Histogram equalized images there is no any sharp patch edges but Stone face is much bright.
Here I have done Non overlapping CLAHE.

Fig1 Fig2

Fig3 Fig4

Fig5
Question2 Image Upsampling
Here I have up sampled two images “Bee Image” and “StoneFace Image” with scaling factor
2 & 3. Using nearest neighbor interpolation and bilinear interpolation.
In first part (Q2a) I have unsampled the image and in Second part(Q2b) I have down
sampled the image and then up sampled the image and calculated mean square error from
both interpolation.
so I have found that Bilinear Interpolation gives better result over Nearest Neighbor
Interpolation because it gives less mean square error.

when We will use scaling factor k =2


➔ Mean Square Error for Bee image using Bilinear interpolation 0.00417

➔ Mean Square Error for StoneFace image using Bilinear interpolation 86.567

➔ Mean Square Error for Bee image using Nearest Neighbour interpolation 0.007959

➔ Mean Square Error for StoneFace image using Nearest Neighbour interpolation 193.77

When We will use scaling factor k =3

➔ Mean Square Error for Bee image using Bilinear interpolation 0.0041725

➔ Mean Square Error for StoneFace image using Bilinear interpolation 86.567146

➔ Mean Square Error for Bee image using Nearest Neighbour interpolation 0.0079592

➔ Mean Square Error for StoneFace image using Nearest Neighbour interpolation 193.7

➔ Mean Square Error for Bee image using Nearest Neighbour interpolation 0.0079
Image Upsampling by factor 2 and 3 Of :
Downsampling and Upsampling of Images
1] Using Bilinear Interpolation
Fig1 Fig2

2] Using Nearest Neighbor Interpolation:


Fig3 Fig4
Question 3: Rotation of Image
Question 4 : Spatial Domain Filtering
4(a) Smoothing:
When Mask size increased images are become more blurred
Fig1: With Mask Size 5 Fig2: With Mask Size 10

Fig1: With Mask Size 5


4(b) High Boost Filtering:
Here First I smooth the noisy image by mask of size 5 filtere . then I have
calculated mean square error between character image and denoised image
from part (a) then I have found factor k for high boost filtering.
Fig1 First step (Noisy Image ) fig2 Second Step (Filtered image with mask 5)

Third Step: finding k by minimum Mean square Error


Fourth Step High Boost Filtering, Sharped Image
Fig3 : Sharped Image
Question4(C):
When we apply high boost filtering to directly noisy image so image will come
more noisy and basically it sharps the high frequency elements.
Noisy Image

After Applying High Boost Filtering

You might also like