This paper presents an evolutionary approach to estimating upper-body posture from multi-view markerless sequences. We fit a 24-dof skeleton model to sparse 3-D stereo data from an array of cameras. We use a particle swarm optimization algorithm which is intrinsically parallel, can incorporate constraints and does not require motion models. We subdivide the high-dimensional search space based on limb dynamics from application sequences and perform hierarchical fitting from the least to the most uncertain body parts. We show experimentally the advantages of this scheme against non-hierarchical optimization in terms of sharper error decrease. We report results with 3-D scanner data of a model human and noisy, calibrated stereo disparity maps of a real videoconferencing scene. 1 Introduction and

In this paper we present an algorithm for automatically building models of articulated objects from range data. These models not only describe the surface shape of the object but also describe the kinematics that constrain the movement of one object component in relation to another. This is more difficult than building models of rigid objects because the association of surface measurements to object components must be determined. The algorithm is demonstrated on a difficult object with free-form surfaces. 1 Introduction The ability to automatically acquire geometric models from example objects is useful in a growing number of application areas. In the field of computer graphics, the need for improvements in realism requires more complex models, but manual model construction is time-consuming and difficult. Users of Computer-Aided Design technology would like to be able to make improvements to a manufactured part and then update their CAD model to reflect this. This provides a very eff...

The paper proposes a reliable method for estimating second-order surfaces from 3D range data in the framework of object recognition and localization or object modelling. Instead of estimating such surface individually the approach ts all the surfaces captured in the scene together, taking into account the geometric relationships between them and their speci c characteristics. The technique is compared with other methods through experiments performed on real objects and demonstrates that the use of constrained relationships improves shape estimates.

Until recently, most real time signal processing has been done in fixed point, due to size, cost, and speed limitations of available hardware. However, the increased dynamic range, consistent precision, and automatic normalization of floating point make it a desirable goal. The same increase in performance that was attained a few years ago when general purpose computing accepted floating point should now be attainable in hardware-based signal processing. For general purpose computation, 32 bit floating point has become the single precision standard. However, the 24 bit significand of this format provides more precision than is needed or justified for most signal processing, requires a large amount of hardware, and cannot yet be realized at 10MHz in a single integrated circuit. Therefore, a compromise format of 22 bits, using a 16 bit significand and a 6 bit exponent, has been developed. This format is compatible with 16 bit fixed point. To support this format, TRW is developing a single chip 10MHz adder/subtractor and a single chip 10MHz multiplier.

Hardware and software needed to implement a 16-bit microprocessor-based, single-board digital signal processing system are outlined and discussed. As described, the system can perform high-speed digital filtering and fast Fourier transforms in real time for signals of up to 35 kHz, and the basic architecture can be extended readily to handle higher frequency signals in real time. The system is built around a 2901B bit-slice microprocessor, two 2911 microcontrollers, a TRW 16-bit multiplier-accumulator, nine 512×8 registered PROMs, and twelve 1024×4 RAMs. Applications of this system to matrix multiplication and other matrix operations are also discussed.