Prediction and Optimization Algorithm for Intersection Point of Spatial Multi-Lines Based on Photogrammetry.

Area of Science:

• Engineering
• Photogrammetry
• Computer Vision

Background:

• Complex engineering environments often feature occluded corners or multi-line intersections, hindering direct measurement.
• Existing photogrammetry techniques face challenges with direct measurement of these occluded spatial intersection points due to factors like image distortion and point selection errors.

Purpose of the Study:

• To propose and validate a novel prediction and optimization algorithm for determining the intersection points of spatial multi-lines using photogrammetry.
• To address the limitations of direct measurement in complex engineering settings by improving the accuracy of intersection point prediction.

Main Methods:

• Utilized photogrammetry to calculate spatial line point coordinates.
• Developed an iterative algorithm to fit initial intersection points using line equations.
• Employed image projection, information entropy, and distance-based weighting to refine intersection point calculations.
• Iteratively re-fitted coordinates until a predefined error threshold or maximum iteration count was met.

Main Results:

• The proposed algorithm demonstrated significant improvements in the prediction accuracy of spatial multi-line intersection points.
• Experimental validation across three distinct scenarios confirmed the algorithm's effectiveness in complex environments.
• The weighting mechanism, incorporating image plane distances and information entropy, proved crucial for accurate refinement.

Conclusions:

• The developed photogrammetry-based algorithm effectively overcomes challenges in measuring occluded spatial intersection points.
• The iterative refinement process significantly enhances the precision of intersection point prediction in engineering applications.
• This approach offers a robust solution for accurate spatial analysis in complex, obstructed environments.