Constraints on deformable models: recovering 3D shape and nongrid motion
Artificial Intelligence
On active contour models and balloons
CVGIP: Image Understanding
Intelligent scissors for image composition
SIGGRAPH '95 Proceedings of the 22nd annual conference on Computer graphics and interactive techniques
Reflectance based object recognition
International Journal of Computer Vision
Color Based Object Recognition
ICIAP '97 Proceedings of the 9th International Conference on Image Analysis and Processing-Volume I - Volume I
Physics-Based Segmentation: Moving Beyond Color
CVPR '96 Proceedings of the 1996 Conference on Computer Vision and Pattern Recognition (CVPR '96)
Shadows and shading flow fields
CVPR '96 Proceedings of the 1996 Conference on Computer Vision and Pattern Recognition (CVPR '96)
| Hi-index | 0.00 |
Interactive segmentation for image manipulation and composition is becoming increasingly important as digital imaging devices become ubiquitous. Users want to select objects in one image to put them in another. This paper proposes a method for selecting image regions that are likely to correspond to multicolored objects rather than just regions of similar color and/or texture. The method is based upon a physical analysis of neighboring regions using the reflectance ratio along the region borders. This measure provides an illumination and geometry invariant cue to scene coherence between regions of different color. Using this measure and a user-selected seed region, the computer can automatically or interactively grow the region to include neighboring regions of different color that are likely to be part of the same object.