Active part-decomposition, shape and motion estimation of articulated objects: a physics-based approach

Abstract

We present a novel, robust, integrated approach to segmentation shape and motion estimation of articulated objects. Initially, we assume the object consists of a single part, and we fit a deformable model to the given data using our physics-based framework. As the object attains new postures, we decide based on certain criteria if and when to replace the initial model with two new models. These criteria are based on the model's state and the given data. We then fit the models to the data using a novel algorithm for assigning forces from the data to the two models, which allows partial overlap between them and determination of joint location. This approach is applied iteratively until all the object's moving parts are identified. Furthermore, we define new global deformations and we demonstrate our technique in a series of experiments, where Kalman filtering is employed to account for noise and occlusion.

Active motion-based segmentation of human body outlines

“…As a first step towards such an algorithm, we have presented a physics-based approach to the shape and motion estimation of non-occluded chain-like structures in human body outlines (e.g., arms and legs) [5]. In this paper, we extend our technique to be able to fully segment outlines of moving humans.…”

“…We initialize a new deformable model ml, and replace mo with the models m(tini,) and ml. We f t these models to the given data, using our weighted-force assignment algorithm [5] and the physics-based framework introduced in [9]. In the following, we describe the modifications to the framework which are needed to accommodate fitting of composed models to time-varying data.…”

“…Later, joints between the parts are estimated employing the algorithm for joint estimation described in [5].…”

“…), where the geometric primitive e (defined parametrically over u with global shape parameters ai) is subjected to the global deformation T which depends on the parameters bi. To represent the local deformations, we employ finite element shape functions which are tensor products of one-dimensional Hermite polynomials [ 5 ] . Therefore, d = S q d , where S is the shape matrix whose entries are the shape functions and q d = (.…”

“…Moreover, no techniques exist that acquire a concise model of the human body automatically. The solution to the above problems requires the development of an algorithm that: (1) integrates the processes of segmentation and fitting, (2) allows reliable shape description of the parts, (3) estimates the true location of the joints between the parts, (4) detects multiple joints, and (5) obviates the need for markers and special equipment.…”