Abstract: A method for detecting an object in an image includes calculating a log L likelihood of pairs of pixels at select positions in the image that are derived from training images. The calculated log L likelihood of the pairs of pixels is compared with a threshold value. The object is detected when the calculated log L likelihood is greater than the threshold value.

Abstract: A method for performing shape localization in an image includes deriving a model shape from a database of a plurality of sample shapes. The model shape is defined by a set of landmarks. The method further includes deriving a texture likelihood model of present sub-patches of the set of landmarks defining the model shape in the image, and proposing a new set of landmarks that approximates a true location of features of the shape based on a sample proposal model of the present sub-patches. A CONDENSATION algorithm is used to derive the texture likelihood model and the proposed new set of landmarks.

Abstract: A method of determining spatial target probability using a model of multisensory processing by the brain includes acquiring at least two inputs from a location in a desired environment where a first target is detected, and applying the inputs to a plurality of model units in a map corresponding to a plurality of locations in the environment. A posterior probability of the first target at each of the model units is approximated, and a model unit with a highest posterior probability is found. A location in the environment corresponding to the model unit with a highest posterior probability is chosen as the location of the next target.

Abstract: A method of determining spatial target probability using a model of multisensory processing by the brain includes acquiring at least two inputs from a location in a desired environment where a first target is detected, and applying the inputs to a plurality of model units in a map corresponding to a plurality of locations in the environment. A posterior probability of the first target at each of the model units is approximated, and a model unit with a highest posterior probability is found. A location in the environment corresponding to the model unit with a highest posterior probability is chosen as the location of the next target.

Abstract: A FAP coding technique that realizes enough coding gain to transmit multiple synthetic talking heads over a band limited channel without introducing perceptible artifacts into the reconstructed synthetic talking heads. This is accomplished by exploiting the spatial correlation of each FAP frame and/or the temporal correlation of the sequence of FAP frames. To remove intra-frame correlation, each FAP frame is transformed prior to segmentation from the n-dimensional space into an m-dimensional subspace where m<n using an energy compaction transform. To remove inter-frame redundancy, the sequence is segmented and each parameter vector is transform coded to decorrelate the vector.

Abstract: Pose estimation of a camera is facilitated by including in the scene a space fiducial, an object that has the property that its appearance as viewed by the camera uniquely determines the direction from which it is viewed.

Abstract: Data which represents an animation sequence of a three dimensional object at a series of discrete time frames is compressed by identifying characteristic features of the object; generating a quadrangular mesh representation of the object, whereby the object is mathematically defined by dividing it into one or more regions and hierarchically representing each region by a mesh, each mesh including three coordinate matrices which define the positions of nodes within the mesh; selecting from the mesh representation of the features a set of animation parameters which are capable of specifying changes in the mesh corresponding to the animation of the object; compressing each region mesh by applying pyramid progressive coding to the coordinate matrices for the mesh; storing the initial values for the animation parameters at the beginning of the animation sequence; and, at each time frame after the beginning of the animation sequence, estimating the current values of the parameters and compressing each parameter by e