Incremental Model Construction using Geometric Constrains
The main motivation of this work is to improve the deformable model shape and motion estimation, in an incremental way, during tracking, by discovering and reconstructing new, independent features whose image motion is consistent with the estimated model motion. Once the features are identifyed, they are reconstructed, integrated into the model and subsequently used to improve tracking robustness (see below for a quantitative plot). The longer-term, higher-level goal is to group them and recover higher level model primitives. The more low-level technical advantage is that while the model coverage in the image is improved during tracking, the estimation is done in the higher-level model representation space, providing compactness and robustness.
Abstract: Previous approaches to deformable model shape estimation and tracking have assumed a fixed class of shapes representation (e.g., deformable superquadrics), initialized prior to tracking. Since the shape coverage of the model is fixed, such approaches do not directly accommodate incremental representation discovery during tracking. As a result, model shape coverage is decoupled from tracking, thereby limiting both processes in terms of scope and robustness. We present a novel deformable model framework that accommodates the incremental incorporation during tracking of new geometric primitives (lines, in addition to points) that are not explicitly captured in the initial deformable model but that are moving consistently with its image motion. As these new features are detected via consistency checks, they are added to the model, providing incremental soft constraints on the estimation of its rigid parameters. The consistency checks are based on trilinear relationships between geometric primitives. Consequently, we not only increase both model scope and, ultimately, its higher-level shape coverage, but improve tracking robustness and accuracy, by directly employing the new features in both forward prediction and reconstruction. Our new formulation is a step towards automating model shape estimation and tracking, since it requires significantly reduced initial model hand-crafting. We demonstrate our approach on two separate image-based tracking domains, each involving complex 3D object shape and motion.
Keywords: Deformable models, geometric constraints, object tracking, model grouping, bundle adjustement.
In the following images (initial sequence courtesy Dr.Piotr Jasiobedzky, MD Robotics) you can see that an initial wireframe (brown) superquadric is fitted to a part of the object and new structure is discovered (green lines) and recovered (red corresponding lines) during tracking via model self-conistency checks. Subsequently, the new recovered structure is used during model motion estimation. Both structure recovery and tracking is robust and un-biased.
