dsef 0.2.0

Direct Step Edge Detection

Direct Step Edge Follower

DSEF Package

The Direct Step Edge Follower (DSEF) is a edge-following algorithm designed for high-precision edge detection with low computational cost. It employs stepwise directional refinement and kernel-based statistical testing to enhance accuracy, particularly in challenging lighting conditions.

Citation

To cite DSEF in your research, please cite as:

@article{Sivertsen2025DSEF,
  author    = {Agnar Sivertsen and Fabio A. A. Andrade and Marcos Moura and Carlos A. M. Correia and Mariane R. Petraglia},
  title     = {Direct Step Edge Follower: a novel edge follower algorithm applied to solar panels inspections with Unmanned Aerial Vehicles},
  journal   = {Preprint},
  year      = {2025},
  month     = {April},
  url       = {https://www.researchgate.net/publication/390370984_Direct_Step_Edge_Follower_a_novel_edge_follower_algorithm_applied_to_solar_panels_inspections_with_Unmanned_Aerial_Vehicles}
}

How to use

Simple Implementation of DSEF Code to search for the closest line in an example image. Full example can be seen here.

Imports

import sys
import os

import cv2
import matplotlib.pyplot as plt

import dsef

Load Image

image_path = 'imgs/input_image.jpg'
if not os.path.isfile(image_path):
    raise FileNotFoundError(f"Image '{image_path}' not found.")

img = cv2.imread(image_path)

# Convert the image from BGR (OpenCV default) to RGB (matplotlib default)
img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)

# Plot the image using matplotlib
plt.imshow(img_rgb)
plt.axis('off')  # Hide the axes
plt.show()

Initiate DSEF Class

initial_direction_deg = 270     # The initial direction in degrees
dir_span = 90                   # The span of directions for search

start_pix = (0+200, 0+200)      # Starting pixel (x, y)
end_pix = (1920, 1080)          # Ending pixel (x, y)
speed = "medium"                # Speed of the algorithm, "high", "medium", or "low"
debug = True

dsef_ = dsef.Dsef(initial_direction_deg, dir_span, start_pix, end_pix, speed, debug)

Edge Search

The dots in the image below represents the steps of the algorithm

## Perform edge search
edge_found, img = dsef_.edge_search(img)

if edge_found:
    print(f"Edge found: {edge_found}")

# Convert the image from BGR (OpenCV default) to RGB (matplotlib default)
img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)

# Plot the image using matplotlib
plt.imshow(img_rgb)
plt.axis('off')  # Hide the axes
plt.show()

[DEBUG] Overriding d_theta to: 4.0
[DEBUG] Distance between start and end: 1932.0455481173315
[DEBUG] Search step: 32.200759135288855
[DEBUG] Follower step: 22.029071700822982
[DEBUG] Edge found. Breaking EdgeSearch.
Edge found: True

Edge Follow

The dots parallel to the edge of the rectangle shows the following path of DSEF algorithm

found_edge_line = None
if edge_found:
    # if the search found an edge, follow it
    found_edge_line, image = dsef_.edge_follow(img)

print(f"Found edge line [(x1,y1),(x2,y2)]: {found_edge_line}")

# Convert the image from BGR (OpenCV default) to RGB (matplotlib default)
img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)

# Plot the image using matplotlib
plt.imshow(img_rgb)
plt.axis('off')  # Hide the axes
plt.show()

[DEBUG] Distance between start and end: 1932.0455481173315
[DEBUG] Search step: 32.200759135288855
[DEBUG] Follower step: 22.029071700822982
[DEBUG] EdgeFollow message: CANCEL. WE LOST THE EDGE
Found edge line [(x1,y1),(x2,y2)]: [(522, 380), (-17152.38437246058, 3496.470832276853)]