Abstract: A method for accelerated decision tree execution in a processor of a digital system is provided that includes receiving at least some attribute values of a plurality of attribute values of a query for the decision tree in a pre-processing component, evaluating the received attribute values in the pre-processing component according to first early termination conditions corresponding to a first decision to determine whether or not the received attribute values fulfill first early termination conditions, and querying the decision tree with the plurality of attribute values when the received attribute values do not fulfill the first early termination conditions.

Abstract: A method for accelerated decision tree execution in a processor of a digital system is provided that includes receiving at least some attribute values of a plurality of attribute values of a query for the decision tree in a pre-processing component, evaluating the received attribute values in the pre-processing component according to first early termination conditions corresponding to a first decision to determine whether or not the received attribute values fulfill first early termination conditions, and querying the decision tree with the plurality of attribute values when the received attribute values do not fulfill the first early termination conditions.

Abstract: A method for accelerated decision tree execution in a processor of a digital system is provided that includes receiving at least some attribute values of a plurality of attribute values of a query for the decision tree in a pre-processing component, evaluating the received attribute values in the pre-processing component according to first early termination conditions corresponding to a first decision to determine whether or not the received attribute values fulfill first early termination conditions, and querying the decision tree with the plurality of attribute values when the received attribute values do not fulfill the first early termination conditions.

Abstract: A method of object retrieval from visual data is provided that includes annotating at least one portion of the visual data with a context keyword corresponding to an object, wherein the annotating is performed responsive to recognition of the context keyword in audio data corresponding to the at least one portion of the visual data, receiving a query to retrieve the object, wherein the query includes a query keyword associated with both the object and the context keyword, identifying the at least one portion of the visual data based on the context keyword, and searching for the object in the at least one portion of the visual data using an appearance model corresponding to the query keyword.

Abstract: A method of camera pose estimation is provided that includes capturing a model image of a scene at a canonical camera pose, generating an image library from warped images of the model image and the model image, wherein each warped image is a transformation of the model image at a different pre-determined camera pose, capturing an image of the scene as a user moves the camera, reporting the current camera pose as a camera pose of the image when the image is acceptable, conditionally adding the first image to the image library when the first image is acceptable, and re-initializing the current camera pose to a camera pose selected from the image library when the first image is not acceptable.

Abstract: A method for three dimensional (3D) image processing is provided that includes receiving an image, wherein each location in the image includes a value indicative of a depth of a pixel in a scene and wherein each value has an associated confidence measure, determining whether each similarity region of a plurality of non-overlapping similarity regions in the image is valid or invalid based on a number of values in the similarity region having sufficiently high confidence measures, wherein a similarity region includes contiguous locations in the image having similar values, and indicating that the values in a similarity region are invalid when the similarity region is determined to be invalid.

Abstract: A method of camera pose estimation is provided that includes capturing a model image of a scene at a canonical camera pose, generating an image library from warped images of the model image and the model image, wherein each warped image is a transformation of the model image at a different pre-determined camera pose, capturing an image of the scene as a user moves the camera, reporting the current camera pose as a camera pose of the image when the image is acceptable, conditionally adding the first image to the image library when the first image is acceptable, and re-initializing the current camera pose to a camera pose selected from the image library when the first image is not acceptable.

Abstract: A method of camera pose estimation is provided that includes capturing a model image of a scene at a canonical camera pose, generating an image library from warped images of the model image and the model image, wherein each warped image is a transformation of the model image at a different pre-determined camera pose, capturing an image of the scene as a user moves the camera, reporting the current camera pose as a camera pose of the image when the image is acceptable, conditionally adding the first image to the image library when the first image is acceptable, and re-initializing the current camera pose to a camera pose selected from the image library when the first image is not acceptable.

Abstract: A method of object retrieval from visual data is provided that includes annotating at least one portion of the visual data with a context keyword corresponding to an object, wherein the annotating is performed responsive to recognition of the context keyword in audio data corresponding to the at least one portion of the visual data, receiving a query to retrieve the object, wherein the query includes a query keyword associated with both the object and the context keyword, identifying the at least one portion of the visual data based on the context keyword, and searching for the object in the at least one portion of the visual data using an appearance model corresponding to the query keyword.

Abstract: A method for approximate pyramidal search for displacement matching is provided that includes performing a complete window-based search at a coarsest resolution level in a resolution pyramid to determine a best matching point in a destination image for each point in a source image, wherein a displacement from each point to the best matching point is determined, counting, for each displacement of the determined displacements, a number of points having the displacement, selecting displacements from the determined displacements based on the number of points counted for each displacement and a predetermined threshold, and performing a search at the next finer resolution level in the resolution pyramid to determine a best matching point in the destination image for each point in the source image, wherein the search for a best matching point for each point is limited to the selected displacements and a reduced search window around each displacement.

Abstract: A method is provided that includes generating coefficients of a scene adaptive filter (SAF) based on differences between values of neighboring pixels in a representative two dimensional (2D) image, and applying the SAF to a plurality of corresponding 2D images.

Abstract: A method for approximate pyramidal search for displacement matching is provided that includes performing a complete window-based search at a coarsest resolution level in a resolution pyramid to determine a best matching point in a destination image for each point in a source image, wherein a displacement from each point to the best matching point is determined, counting, for each displacement of the determined displacements, a number of points having the displacement, selecting displacements from the determined displacements based on the number of points counted for each displacement and a predetermined threshold, and performing a search at the next finer resolution level in the resolution pyramid to determine a best matching point in the destination image for each point in the source image, wherein the search for a best matching point for each point is limited to the selected displacements and a reduced search window around each displacement.

Abstract: A method of camera pose estimation is provided that includes capturing a model image of a scene at a canonical camera pose, generating an image library from warped images of the model image and the model image, wherein each warped image is a transformation of the model image at a different pre-determined camera pose, capturing an image of the scene as a user moves the camera, reporting the current camera pose as a camera pose of the image when the image is acceptable, conditionally adding the first image to the image library when the first image is acceptable, and re-initializing the current camera pose to a camera pose selected from the image library when the first image is not acceptable.