A Stereo Dual-Channel Dynamic Programming Algorithm for UAV Image Stitching

Area of Science:

• Computer Vision
• Robotics
• Geospatial Analysis

Background:

• Dislocation presents a significant challenge in stitching images captured by unmanned aerial vehicles (UAVs).
• Existing methods often struggle with accuracy and robustness, especially in complex scenarios like dense urban areas.
• The classical Duplaquet algorithm has theoretical limitations that hinder optimal seam line selection.

Purpose of the Study:

• To develop a novel algorithm for seamless UAV image stitching.
• To address and overcome the dislocation problem in UAV image stitching.
• To improve the adaptability and robustness of image stitching algorithms.

Main Methods:

• A two-step approach involving image matching for coordinate system correction.
• A new dynamic programming algorithm utilizing stereo dual-channel energy accumulation.
• An innovative energy aggregation and traversal strategy for optimal seam line identification.

Main Results:

• The proposed algorithm effectively resolves dislocation issues in UAV image stitching.
• Demonstrated superior performance, especially in dense urban environments.
• The solution is direction-independent, enhancing adaptability and robustness.

Conclusions:

• The dynamic programming approach offers a robust solution for UAV image stitching challenges.
• The algorithm provides more optimal seam lines compared to classical methods.
• This method significantly improves the quality and reliability of stitched UAV imagery.