🖼️ imagex

This project is a MATLAB implementation of a simple image processing manager for applying 3x3 convolution kernels (filters) to RGB images. It includes a core class (ImageEnhancer), a manual convolution function, a non-interactive demonstration script, and a basic Graphical User Interface (GUI).

The primary goal is to demonstrate Object-Oriented Programming (OOP) principles in MATLAB while providing a hands-on example of 2D image convolution with various boundary padding modes.

---

🚀 Getting Started

Prerequisites

• MATLAB (R2018a or newer is recommended for full GUI functionality).
• The project assumes an image file named peppers.png exists in the directory for the main.m demonstration script to run successfully. This image is commonly included with MATLAB.

Running the Code

1. Download: Clone or download all the files into a single directory.
2. Set Path: Open MATLAB and set the current folder to the project directory.
3. Run Demos:
   • Interactive GUI: Type mainUI in the MATLAB Command Window. This allows you to upload an image, select a kernel, and choose a processing mode (RGB or single channel).
   • Scripted Demo: Type main in the MATLAB Command Window. This runs a non-interactive script that compares several filters on a grayscale image and demonstrates RGB sharpening.

---

📁 Project Structure

File	Description
ImageEnhancer.m	Core Class. Manages image loading, normalization, kernel definitions, and coordinates the filtering of individual channels.
conv2d_manual.m	Core Function. Implements the 2D convolution algorithm using nested loops, supporting custom boundary padding.
main.m	Scripted Demonstration. Runs several examples using the ImageEnhancer class and conv2d_manual for comparison.
mainUI.m	Graphical User Interface (GUI). Provides an interactive way to test the filters and modes.

---

🔧 Core Components Details

ImageEnhancer Class

The class constructor accepts either an image file path (string) or a numeric HxWx3 image matrix. It normalizes all inputs to an uint8 HxWx3 format.

Defined Kernels

The class includes a static method defineKernels() with the following pre-defined 3x3 filters:

• Identity
• Box (Averaging/Blurring)
• Sharpen
• Gaussian
• SobelV (Vertical Edge Detection)
• SobelH (Horizontal Edge Detection)
• Laplacian (Edge Detection)
• Emboss

Processing Methods

• processChannel(channelChar, kernelNameOrMatrix): Filters a single channel ('R', 'G', or 'B').
• processRGB(kernelNameOrMatrix): Filters all three channels independently and combines the results.

conv2d_manual Function

This function handles the low-level mechanics of convolution. It converts the input channel to double for computation and flips the kernel 180 degrees to conform to the mathematical definition of convolution.

Supported Padding Modes

The function supports three boundary handling strategies for a 3x3 kernel (1-pixel padding):

1. zero: Fills the border with zeros (default for standard signal processing).
2. replicate: Copies the edge pixels outward.
3. reflect: Mirrors the internal pixels across the boundary (symmetric padding).