What question did this study set out to answer?

Develop and evaluate an autonomous image filtering and segmentation framework using the nonparametric mean shift density gradient estimator in the joint spatial-range domain.

January 1, 1999

Mean shift analysis and applications

View Full Paper

Ask AI

Bookmark

Authors

DCDorin Comaniciu PMPeter Meer

Discussion

Member takes

Overview

Computational study demonstrates autonomous discontinuity-preserving filtering and segmentation in digital images, indicating robust edge-preserving image processing without manual intervention.

Key Points

Develop and evaluate an autonomous image filtering and segmentation framework using the nonparametric mean shift density gradient estimator in the joint spatial-range domain.
Applied a nonparametric density gradient estimator (mean shift) in the joint spatial-range domain of grayscale and color images to associate pixels with local density modes.
Proved mathematical convergence of the mean shift algorithm on discrete image lattices.
Constructed a segmentation framework by fusing adjacent regions associated with nearby modes using two resolution-control parameters.
Formally established guaranteed convergence of the mean shift procedure on discrete lattices, removing the need for manual stopping criteria.
Achieved discontinuity-preserving filtering and piecewise constant segmentation across gray and color test images with performance comparing favorably to existing literature methods.

Cite This Study

Comaniciu et al. (1999) studied this question.

synapsesocial.com/papers/6a109aebd478ddac0ffd436d https://doi.org/10.1109/iccv.1999.790416

View Full Paper

Ask AI

Bookmark

January 1, 1999

Mean shift analysis and applications

View Full Paper

Ask AI

Bookmark

Authors

DCDorin Comaniciu PMPeter Meer

Discussion

Member takes

Overview

Computational study demonstrates autonomous discontinuity-preserving filtering and segmentation in digital images, indicating robust edge-preserving image processing without manual intervention.

Key Points

Develop and evaluate an autonomous image filtering and segmentation framework using the nonparametric mean shift density gradient estimator in the joint spatial-range domain.
Applied a nonparametric density gradient estimator (mean shift) in the joint spatial-range domain of grayscale and color images to associate pixels with local density modes.
Proved mathematical convergence of the mean shift algorithm on discrete image lattices.
Constructed a segmentation framework by fusing adjacent regions associated with nearby modes using two resolution-control parameters.
Formally established guaranteed convergence of the mean shift procedure on discrete lattices, removing the need for manual stopping criteria.
Achieved discontinuity-preserving filtering and piecewise constant segmentation across gray and color test images with performance comparing favorably to existing literature methods.

Cite This Study

Comaniciu et al. (1999) studied this question.

synapsesocial.com/papers/6a109aebd478ddac0ffd436d https://doi.org/10.1109/iccv.1999.790416

View Full Paper

Ask AI

Bookmark