Digital Image Processing

Overview

This project provides functionality to enhance or transform images through various filters and transformations, focusing on tasks like noise reduction, edge detection, and intensity adjustment. Below is a detailed explanation of the components, filters, and functionality.

---

Features

1. Upload and Display:
   • Upload an image through the interface.
   • View the original image in a designated canvas.
2. Filters:
   • Apply a variety of filters to enhance or transform the image.
3. Before and After Canvases:
   • Compare the original image and the processed image side by side.
4. Histogram:
   • Display the image histogram to analyze pixel intensity distribution.
5. Buttons:
   • Reset: Reverts the image to its original state.
   • Upload: Uploads a new image.
   • Apply: Applies the selected filter to the image.
   • Save: Saves the processed image to the disk.
   • Select Filter: Dropdown or selection to choose the desired filter.
   • Conditional Rendering:
     • K-Means: Option to apply K-Means clustering.
     • Periodic Noise Direction: Option to select the direction for periodic noise filtering.

---

Filters

1. Median Filter

• Description: Reduces noise by replacing each pixel value with the median value of its neighbors. Useful for noise reduction, particularly salt-and-pepper noise.

  

2. Averaging Filter

• Description: Smoothens the image by averaging pixel values in a kernel. Effective for blurring and reducing random noise.

  

3. Low-Pass Filters

• Description: Removes high-frequency noise while preserving low-frequency details. Ideal for noise suppression and overall smoothing.

  

4. Canny Edge Detection

• Description: Detects edges in an image by finding areas of rapid intensity change. Suitable for edge detection tasks.

  

5. Periodic Noise Filter

• Description: Removes periodic noise patterns from the image. Useful for noise reduction in images with repeating patterns of distortion.

  

6. RGB to Grayscale

• Description: Converts the image from RGB color space to grayscale. Often used for simplifying images by reducing them to intensity-based representations.

  

---

Model

K-Means Clustering

• Description: Groups similar pixel values into clusters to simplify and segment images. In this project, it is designed to work with datasets consisting of numbers between 0 and 9, and the algorithm is responsible for predicting the numbers based on clustering.

  

---

Interface Components

Buttons:

1. Reset:
   • Clears all applied filters and resets the image to its original state.
2. Upload:
   • Opens a file dialog to upload a new image.
3. Apply:
   • Executes the selected filter on the uploaded image.
4. Save:
   • Saves the processed image in the desired format.
5. Select Filter:
   • Dropdown or radio buttons to choose the filter to be applied.

Canvases:

1. Before Canvas:
   • Displays the original image.
2. After Canvas:
   • Displays the processed image.
3. Histogram Canvas:
   • Plots the histogram of the image for pixel intensity analysis.

Conditional Rendering:

• K-Means Clustering:
  • Allows the user to cluster image colors into groups for classification (unsupervised).
• Periodic Noise Direction:
  • Provides options to select the direction for periodic noise filtering.

---

Workflow

1. Upload an Image:
   • Use the "Upload" button to load an image into the application.
2. Select a Filter:
   • Choose the desired filter from the "Select Filter" dropdown.
3. Apply the Filter:
   • Click the "Apply" button to process the image with the selected filter.
4. View Results:
   • Compare the original and processed images in the "Before" and "After" canvases.
   • View the histogram for further analysis.
5. Save the Processed Image:
   • Use the "Save" button to export the processed image.
6. Reset (if needed):
   • Click the "Reset" button to start over.
   • Delete histogram from your device.

---

Additional Notes

• Ensure that the required Python packages are installed:

  pip install numpy==1.21.0 pillow opencv-python matplotlib==3.4.3 scikit-image

• Periodic noise filtering requires specifying a direction or type.

• The application supports both grayscale and color images.