$(12$ Marks$)$

I. Data Collection: Discuss the Source images collected for the

experimentation, e$.$g benchmark images or real images

Demonstrates how to restore the images by using spatial and

Frequency domain filters. The goal is to remove the noise present in

the image. Perform the below steps

a$.$ Read the images files using OpenCV library.

b$.$ Preprocess the dataset images using Median, Gaussian filter and

Laplacian filters in spatial domain to obtain smooth images.

c$.$ Preprocess the images in Frequency domain by DFT or FFT to

images, remove the noise from the images using low pass and

high pass filters.

d$.$ Plot the original image and filtered images.

e$.$ Evaluate the images using MSE or PSNR between the filtered

and original images.

f$.$ Justify your results which filters works better.

II$.$ Demonstrates the process of Object detection and segmentation, for

the de$-$noised images of the Step $1,$ perform the below steps.

a$.$ The first step in this process is to load a preprocessed image from

Phase$1.$

b$.$ Detect the corners using Haris's corner detection.

c$.$ Extract Features by applying EM or Mean shift or watershed

segmentation techniques

d$.$ The next step is to create a structural element, or kernel that

resembles the objects present in the image.

e$.$ Apply erosion or dilation based on the application and display the

contents on the image.

f$.$ The next step is load an image consisting of human faces, apply

Viola Jones Algorithm to detect Faces.

g$.$ Plot the results for all the above steps using OpenCV Plots and

graphs, discuss the improvement in performance