Edge Detection

This repository contains a program that implements three methods for global edge detection, specifically designed for the digital image processing discipline. The base algorithms used in this program are comprehensively explained in the Segmentacao de Imagens.pdf file included in the repository. Specifically:

• The Local Edge Detection algorithm is detailed on page 42.
• The Regional Edge Detection algorithm is discussed on page 47.
• The Global Edge Detection algorithm is explained on page 54.

This documentation provides in-depth insights and theoretical background on the implementation and functioning of each algorithm.

Overview

The program processes images using three different algorithms for edge detection:

1. Local Edge Detection
2. Regional Edge Detection
3. Global Edge Detection

Each algorithm can be customized using various parameters to achieve the desired edge detection results.

Features

• Local Edge Detection: Utilizes magnitude and angle thresholds for precise edge detection.
• Regional Edge Detection: Applies a threshold to the entire image for regional edge detection.
• Global Edge Detection: Implements the Hough Transform to detect edges globally, with options for peak detection, continuous lines, and using an empty image as the base.

Arguments

• <filename>: Path to the image file.
• <algorithm>: The edge detection algorithm to use (local, regional, global).

Options

• --threshold: Threshold for Regional Edge Detection (default: 100).
• --magnitude_threshold: Magnitude for Local Edge Detection (default: 100).
• --angle: Angle for Local Edge Detection (default: "all").
• --angular_threshold: Threshold of Angle for Local Edge Detection (default: 20).
• --reconstruction_size: Max pixels for edge reconstruction in Local Edge Detection (default: 5).
• --gap: Gap used in Global Edge Detection to define the maximum distance between two points to be considered continuous (default: 10).
• --peaks_amount: Amount of peaks to be detected in Global Edge Detection (default: 5).
• --save: Path to save the output image.
• --continuous_lines: Use continuous lines in the output for Global Edge Detection (default: False).
• --empty_image: Use an empty image as the base for the output in Global Edge Detection (default: False).

Usage

Command Line Interface

The program provides a command-line interface (CLI) for ease of use. Below is the main structure of the command to run the program:

Local

 python main.py images/tijolo.png local

 python main.py images/tijolo.png local --angle 90 --magnitude_threshold 150

Regional

Global

python3 main.py images/global/shapes_pp_canny.png global --peaks 12 --save images/global/output_shapes.png

python3 main.py images/global/shapes_pp_canny.png global --peaks 2 --empty_image --continuous_lines --save images/global/output_shapes.png

Preprocessing Input Images

The expected input image for the edge detection program should be preprocessed. To facilitate this, we provide a preprocessing script, canny.py, which applies Canny edge detection to the input image.

Canny Edge Detection Preprocessing

The canny.py script processes an image using the Canny edge detection method and saves the preprocessed image for further use in the main edge detection program.

Usage

Command Line Interface

The script provides a command-line interface (CLI) for ease of use. Below is the main structure of the command to run the preprocessing script:

python canny.py <filename> [options]

Arguments

• <filename>: Path to the image file.

Options

• --thresholds: Thresholds for Canny edge detection (low, high) (default: [50, 150]).