Adaptive deformation correction of depth from defocus for object reconstruction

Area of Science:

• Computer Vision
• Optical Engineering
• Image Processing

Background:

• Three-dimensional (3D) object reconstruction accuracy using depth from defocus (DfD) is crucial in various applications.
• Elliptical lens deformation is a significant factor that degrades the precision of DfD-based 3D reconstruction.
• Existing DfD methods struggle with lens-induced distortions, necessitating robust correction techniques.

Purpose of the Study:

• To develop and evaluate novel methods for correcting elliptical lens deformation in depth from defocus.
• To enhance the accuracy and reliability of 3D object reconstruction in the presence of lens aberrations.
• To address the low-texture problem often encountered in DfD techniques.

Main Methods:

• Correction by Deformation Cancellation (CDC): Subtracts current deformed depth values from prestored deformed values.
• Correction by Least Squares Fit (CLSF): Maps observed deformed depth values to expected, undistorted values.
• Integration of smoothing algorithms post-correction to mitigate issues arising from low-texture surfaces in DfD.

Main Results:

• Both CDC and CLSF methods demonstrated significant effectiveness in eliminating deformation-induced errors.
• The proposed correction methods were tested across four different DfD techniques, showing consistent improvements.
• The combined approach of correction and smoothing successfully addressed the low-texture problem, enhancing overall reconstruction quality.

Conclusions:

• The presented CDC and CLSF methods provide efficient and effective solutions for correcting elliptical lens deformation in DfD.
• These techniques substantially improve the accuracy of 3D object reconstruction, making DfD more robust.
• The findings contribute to advancing the practical application of depth from defocus in real-world imaging scenarios.