Analytical Assessment of Image Registration Accuracy after Projective Transformation Compensation

Analytical Assessment of Image Registration Accuracy after Projective Transformation Compensation

Khodakow N.A. (RSREU, Ryazan, Russia)
Babayan P.V. (RSREU, Ryazan, Russia)

Abstract

The problem of analytical assessment of image registration accuracy for projective transformations is addressed. A mathematical framework for computing the covariance matrix of transformation parameters based on linearization of the nonlinear model using Jacobian is proposed. An algorithm for constructing variance maps D_x, D_y and covariance K_xy for each image pixel is developed. Method validation using Monte Carlo numerical approach demonstrated high accuracy of analytical estimates (correlation > 0.999) and significant computational speedup (more than 5000 times). The scientific novelty lies in developing an analytical method for assessing local registration accuracy for nonlinear projective transformations.

Keywords

image alignment; reference points; geometric transformations; image stabilization; projective transformation; correlation matrix; Monte Carlo method.

Edition

Proceedings of the Institute for System Programming, vol. 38, issue 3, part 3, 2026, pp. 125-134

ISSN 2220-6426 (Online), ISSN 2079-8156 (Print).

DOI: 10.15514/ISPRAS-2026-38(3)-40

For citation

Khodakow N.A., Babayan P.V. Analytical Assessment of Image Registration Accuracy after Projective Transformation Compensation. Proceedings of the Institute for System Programming, vol. 38, issue 3, part 3, 2026, pp. 125-134 DOI: 10.15514/ISPRAS-2026-38(3)-40.

Full text of the paper in pdf (in Russian) Back to the contents of the volume