🛠️ Image Preprocessing Specialist Tool

---

This project provides a comprehensive web tool for image preprocessing, built with Python, OpenCV, and NumPy, and deployed using Streamlit. It allows users to visually analyze the effects of various image manipulation and computer vision algorithms.

Use images in "sample_images" folder to try out our tool!

✨ Features and Algorithms

The tool is equipped with methods across three main categories: Noise Generation, Denoising (Filtering), and Feature Detection (Edge Detection).

1. 📢 Noise Generation

Method	Description	Key Principle
Gaussian Noise	Distributes pixel intensity values randomly according to a Gaussian distribution, simulating sensor noise.	Uses np.random to add general random intensity variation.
Salt and Pepper Noise	Randomly introduces isolated bright (white) and dark (black) pixels across the image.	Uses np.random to target specific pixels for extreme value assignment.

2. 🧹 Denoising Filters

Method	Description	Best for Filtering...	OpenCV Function
Gaussian Blur	Applies a weighted average across a pixel neighborhood to smooth the image.	Gaussian Noise	cv2.GaussianBlur()
Median Blur	Replaces the center pixel value with the statistical median value in its neighborhood.	Salt and Pepper Noise	cv2.medianBlur()

3. 🖼️ Image Enhancement and Edge Detection

Method	Description	Core Implementation Steps
Histogram Equalization	Redistributes the image's pixel intensity values to cover the full dynamic range, enhancing global contrast.	1. Converts to YUV. 2. Equalizes the Luminance (Y) channel. 3. Converts back to BGR.
Difference of Gaussians (DoG)	A feature enhancement technique that finds details existing at a specific scale by comparing two blurred versions.	1. Grayscaling. 2. Apply two Gaussian Blurs ($\sigma_1$ and $\sigma_2$). 3. Subtract $\text{Blur}_2$ from $\text{Blur}_1$.
Canny Edge Detection	A robust, multi-stage algorithm designed to find optimal, thin, continuous edges.	1. Noise Reduction (Gaussian Blur). 2. Gradient Calculation (Sobel filters). 3. Non-Max Suppression (Thinning edges). 4. Double Thresholding. 5. Hysteresis Tracking (Connecting weak edges to strong ones).

---

💻 Technical Stack

• Core Logic: imgprocess_fx.py (Contains all custom function implementations).
• Libraries:
  • OpenCV (cv2): Primary library for image loading, filtering, and color space conversion.
  • NumPy: Used for efficient mathematical operations and array manipulation (image representation).
  • Streamlit: Used to create the interactive, user-friendly web interface.