University of Washington

Todd Schoepflin

Yongmin Kim

David R Haynor

Vikram Chalana

Shijun Sun

John D Sahr

HyunWook Park

Control points used in deriving an object boundary for a prior frame are overlaid onto a current frame. An initial estimate of an object boundary are derived from the control points and edge energy data. The operator adjusts the control points to better model the boundary for the current frame. For each updated control point, the object boundary is rederived. A restricted area is defined encompassing the initial control points. When a control point is moved outside the restricted area, the restricted area is redefined to accommodate it. The boundary between control points is derived by finding a best path. Only points within the restricted area are considered. A first set of rules is used to find the best path when the distance between the two points is less than threshold value. A second set of rules is used when the distance between the two points exceeds the threshold value.

Interactive frame segmentation with dynamic programming

Object segmentation and tracking are improved by including directional information to guide the placement of an active contour (i.e., the elastic curve or 'snake') in estimating the object boundary. In estimating an object boundary the active contour deforms from an initial shape to adjust to image features using an energy minimizing function. The function is guided by external constraint forces and image forces to achieve a minimal total energy of the active contour. Both gradient strength and gradient direction of the image are analyzed in minimizing contour energy for an active contour model.

Video object segmentation using active contour model with directional information

An object is tracked among a plurality of image frames. In an initial frame an operator selects an object. The object is distinguished from the remaining background portion of the image to yield a background and a foreground. A model of the background is used and updated in subsequent frames. A model of the foreground is used and updated in the subsequent frames. Pixels in subsequent frames are classified as belonging to the background or the foreground. In subsequent frames, decisions are made, including: which pixels do not belong to the background; which pixels in the foreground are to be updated; which pixels in the background were observed incorrectly in the current frame; and which background pixels are being observed for the first time. In addition, mask filtering is performed to correct errors, eliminate small islands and maintain spatial and temporal coherency of a foreground mask.

Video object tracking by estimating and subtracting background

An object mask and a set of control points defined for a given frame are subjected to morphological processing to remove false edge points and provide a more robust mask for use in tracking the object in a next frame. The morphological processing is performed on a frame by frame basis corresponding to object tracking so that errors added in by the object tracker do not accumulate, and instead are filtered out. Rapidly moving objects which are troublesome for edge-based object trackers are more readily tracked. Also, regions of low contrast or regions locked onto in the background when trying to identify the object are more readily distinguished and eliminated from the object mask.

Morphological postprocessing for object tracking and segmentation

A hierarchy of deformation operations is implemented to deform a template and match the deformed template to an object in a video frame. At each level, the constraints on the template deformations are relaxed, while the spatial range of the object boundary search is narrowed. At a highest level, an initial template is translated, rotated and scaled to coarsely locate the object within a given image frame. At a middle level, an affine transformation is implemented, globally or locally, to deform the template. For a local affine transformation process, a sup-portion, such as an articulation or appendage portion of the template is deformed. The middle level refines the template to get the template boundary close to the actual object boundary within a given frame. At the lowest level, a local segmentation algorithm is applied to deform the now close boundary to finely match the object boundary.

Growing community of inventors

Seattle, WA, United States of America

Todd Schoepflin