Large Displacement Detection Using Improved Lucas-Kanade Optical Flow

Area of Science:

• Civil Engineering
• Computer Vision
• Structural Health Monitoring

Background:

• Structural displacement evaluation is crucial for civil engineering safety.
• Traditional methods for monitoring structural displacement have limitations.
• Lucas-Kanade (LK) optical flow is effective for small displacements but not large ones.

Purpose of the Study:

• To develop an upgraded LK optical flow method capable of detecting large displacement motions.
• To verify the enhanced method's accuracy using controlled and free-falling experiments.
• To compare the performance of the upgraded method with existing techniques like template matching.

Main Methods:

• An upgraded Lucas-Kanade (LK) optical flow algorithm was developed.
• Experiments included motion controlled by a multiple purpose testing system (MTS) and a free-falling setup.
• The upgraded method incorporated pyramid and warp optical flow techniques with a second derivative Sobel operator for free-fall analysis.

Main Results:

• The upgraded LK optical flow method demonstrated 97% accuracy in tracking MTS piston movement.
• For free-falling large displacements, the warping algorithm achieved an average accuracy of 96%.
• The enhanced method successfully captured large displacement motions, outperforming limitations of standard LK optical flow.

Conclusions:

• The upgraded LK optical flow method is a reliable tool for monitoring large structural displacements.
• The integration of pyramid and warp methods enhances accuracy in complex scenarios.
• This advancement offers improved structural health monitoring capabilities for civil engineering applications.