US 8,045,805 B2 · Method for determining whether a feature of interest or an anomaly is present in an image

Abstract

A system and method for identifying objects of interest in image data is provided. The present invention utilizes principles of Iterative Transformational Divergence in which objects in images, when subjected to special transformations, will exhibit radically different responses based on the physical, chemical, or numerical properties of the object or its representation (such as images), combined with machine learning capabilities. Using the system and methods of the present invention, certain objects that appear indistinguishable from other objects to the eye or computer recognition systems, or are otherwise almost identical, generate radically different and statistically significant differences in the image describers (metrics) that can be easily measured.

Claims36

Dependent claims

A processor-implemented method of determining whether a feature of interest is present, comprising:

receiving an image;
mapping the image to a template using at least one predetermined nonlinear divergence transform to yield a mapped image, wherein the at least one predetermined nonlinear divergence transform comprises a transfer function with at least one nodal point positioned so as to cause the feature of interest to exhibit a characteristic color space signature;
and
determining whether the characteristic color space signature is present in the mapped image;
wherein the at least one predetermined nonlinear divergence transform is determined by applying at least one initial divergence transform to an image containing a known feature that is substantially similar to the feature of interest, and adjusting the at least one initial divergence transform so that the known feature diverges from other features in the image by exhibiting the characteristic color space signature.

2 more elements

A processor-implemented method of determining whether an object of interest is present, comprising:

receiving an image;
processing the image, wherein the processing comprises applying at least one transform that bifurcates objects that match predetermined color, density and/or effective atomic number characteristics, wherein the at least one transform comprises at least a first predetermined nonlinear divergence transform, wherein the at least first predetermined nonlinear divergence transform is determined by applying at least one initial divergence transform to an image containing a known object that is substantially similar to the object of interest, and adjusting the at least one initial divergence transform so that the known object is bifurcated from other objects in the image;
isolating possible objects of interest based on the response to the at least one transform;
subjecting possible objects of interest to a feature extraction process that comprises subjecting original pixels that make up the possible objects of interest to at least one transform, wherein the at least one transform comprises at least one other predetermined nonlinear divergence transform, wherein the at least one other predetermined nonlinear divergence transform is determined by applying at least one initial divergence transform to an image containing a known object that is substantially similar to the object of interest, and adjusting the at least one initial divergence transform so that the known object is bifurcated from other objects in the image;
and
determining whether the image contains the object of interest based on the results of the feature extraction process.

3 more elements

Dependencies are inferred from the claim text.

Method for determining whether a feature of interest or an anomaly is present in an image

Life of this patent

Method for determining whether a feature of interest or an anomaly is present in an image

Life of this patent

Abstract

Claims36