MATLAB Implementation of Roberts Operator for Edge Detection

This document presents implementations of several key edge detection operators including Roberts, Prewitt, Sobel, Marr (Laplacian of Gaussian), and Canny operators for processing input images and generating edge-detected outputs. Each operator is implemented through custom MATLAB code that defines appropriate convolution kernels and applies thresholding techniques to identify image edges. The Roberts operator specifically uses 2x2 convolution kernels for diagonal gradient detection, while larger kernels (3x3 for Prewitt/Sobel) capture horizontal and vertical edges with different noise sensitivity characteristics. The implementation includes proper image preprocessing steps such as grayscale conversion and noise reduction using Gaussian filtering for the Marr and Canny operators. The results are systematically compared with MATLAB's built-in edge detection functions (like edge() with various method parameters) through quantitative metrics including edge connectivity analysis and noise robustness evaluation. This comparative approach provides deeper insights into edge detection algorithm performance and helps understand the practical trade-offs between computational efficiency and detection accuracy in image processing applications.