Abstract: An image comparison method constituted of: receiving images; point matching the images to identify tentatively corresponding points; responsive to the identified points, defining a plurality of first image tiles within the first image; defining a plurality of second image tiles within the second image, each corresponding to a respective first image tile; adjusting the intensity of a set of pixels responsive to the cumulative relative frequency of the respective pixel intensity within the respective tile; for each tile, applying a plurality of non-linear functions; for each function, separately determining a moment of the outcome of the respective non-linear function for each axis of the respective tile; for each first image tile and corresponding second image tile, determining a distance between the subspaces spanned by moment vectors; and responsive to determined distances, determining whether respective first and second image tiles comprise observations of the same portion.

Abstract: Methods of creating frames captured with a video camera that are independent of light conditions are presented. In a first method the invention attains this goal by the addition of a modulated light source, for example amplitude modulation (AM), to the background light that illuminates the scene being captured and executing an algorithm that generates processed images in which the effects of the background illumination have been eliminated. In a second aspect the invention is a method of modifying the flash no flash technique by executing an algorithm that averages out the maximum and minimum intensity peaks in adjacent frames thereby allowing this technique to be employed using non-synchronous light in video.

Abstract: Methods of creating frames captured with a video camera that are independent of light conditions are presented. In a first method the invention attains this goal by the addition of a modulated light source, for example amplitude modulation (AM), to the background light that illuminates the scene being captured and executing an algorithm that generates processed images in which the effects of the background illumination have been eliminated. In a second aspect the invention is a method of modifying the flash no flash technique by executing an algorithm that averages out the maximum and minimum intensity peaks in adjacent frames thereby allowing this technique to be employed using non-synchronous light in video.

Abstract: An image comparison method constituted of: receiving images; point matching the images to identify tentatively corresponding points; responsive to the identified points, defining a plurality of first image tiles within the first image; defining a plurality of second image tiles within the second image, each corresponding to a respective first image tile; adjusting the intensity of a set of pixels responsive to the cumulative relative frequency of the respective pixel intensity within the respective tile; for each tile, applying a plurality of non-linear functions; for each function, separately determining a moment of the outcome of the respective non-linear function for each axis of the respective tile; for each first image tile and corresponding second image tile, determining a distance between the subspaces spanned by moment vectors; and responsive to determined distances, determining whether respective first and second image tiles comprise observations of the same portion.

Abstract: Methods for estimating joint geometric and radiometric deformations relating two observations of the same object provide methods for identifying and matching images (as in face recognition) and for characterizing and determining the relationship between a pair of observations (as in target recognition).

Abstract: In object registration and recognition based on a set of known templates, the tremendous set of possible transformations that may relate the template and an observed signature makes any detection and recognition problem ill-defined unless this variability is taken into account. The present invention estimates the deformation that transforms some pre-chosen representation of an object (template) into the current observation. The method employs a set of non-linear operators to replace a high dimensional problem by an equivalent linear problem, expressed in terms of the unknown parameters of the transformation model. The solution is applicable to any homeomorphic transformation regardless of its magnitude. In the special case where the transformation is affine the solution is shown to be exact.

Abstract: In object registration and recognition based on a set of known templates, the tremendous set of possible transformations that may relate the template and an observed signature makes any detection and recognition problem ill-defined unless this variability is taken into account. The present invention estimates the deformation that transforms some pre-chosen representation of an object (template) into the current observation. The method employs a set of non-linear operators to replace a high dimensional problem by an equivalent linear problem, expressed in terms of the unknown parameters of the transformation model. The solution is applicable to any homeomorphic transformation regardless of its magnitude. In the special case where the transformation is affine the solution is shown to be exact.

Abstract: In object registration and recognition based on a set of known templates, the tremendous set of possible transformations that may relate the template and an observed signature makes any detection and recognition problem ill-defined unless this variability is taken into account. The present invention estimates the deformation that transforms some pre-chosen representation of an object (template) into the current observation. The method employs a set of non-linear operators to replace a high dimensional problem by an equivalent linear problem, expressed in terms of the unknown parameters of the transformation model. The solution is applicable to any homeomorphic transformation regardless of its magnitude. In the special case where the transformation is affine the solution is shown to be exact.

Abstract: In object registration and recognition based on a set of known templates, the tremendous set of possible transformations that may relate the template and an observed signature makes any detection and recognition problem ill-defined unless this variability is taken into account. The present invention estimates the deformation that transforms some pre-chosen representation of an object (template) into the current observation. The method employs a set of non-linear operators to replace a high dimensional problem by an equivalent linear problem, expressed in terms of the unknown parameters of the transformation model. The solution is applicable to any homeomorphic transformation regardless of its magnitude. In the special case where the transformation is affine the solution is shown to be exact.