Methods, apparatuses, and computer-readable media for projectional morphological analysis of N-dimensional signals

Patent metadata

Field	Value
Publication number	US-9342876-B2
Application number	US-201313870700-A
Country	US
Kind code	B2
Filing date	Apr 25, 2013
Priority date	Apr 25, 2013
Publication date	May 17, 2016
Grant date	May 17, 2016

Abstract

Embodiments discussed herein in the form of methods, systems, and computer-readable media deal with the application of advanced “projectional” morphological algorithms for solving a broad range of problems. In a method of performing projectional morphological analysis, an N-dimensional input signal is supplied. At least one N-dimensional form indicative of at least one feature in the N-dimensional input signal is identified. The N-dimensional input signal is filtered relative to the at least one N-dimensional form and an N-dimensional output signal is generated indicating results of the filtering at least as differences in the N-dimensional input signal relative to the at least one N-dimensional form.

First claim

What is claimed is:

1. A method of performing projectional morphological analysis, the method comprising:
supplying a first N-dimensional input image (f) representative of a scene of a field of view;
defining non-intersecting subsets (A) from the first N-dimensional input image (f) in which points have a constant brightness value (C) within its respective subset (A), a union of the non-intersecting subsets (A) denoted as set X;
supplying a second N-dimensional input image (g) representative of the scene of the field of view taken under different conditions, the second N-dimensional input image (g) also defined by the set X having the same defined non-intersecting subsets (A) to denote an N-dimensional form (V g ) of the second N-dimensional input image (g);
computing an average brightness for each subset (A) of the N-dimensional form (V g ) to generate a morphological projection image (g′); and
generating an N-dimensional output signal indicating a morphological difference of the first N-dimensional input image (f) and the second N-dimensional input image (g) by subtracting morphological projection image (g′) from the second N-dimensional input image (g).

2. The method of claim 1 , wherein the first N-dimensional input image (f) is a 2-dimensional image, the second N-dimensional input image (g) is a 2-dimensional image, the N-dimensional form (V g ) is a 2-dimensional image, and the N-dimensional output signal is a 2-dimensional image.

3. The method of claim 1 , wherein the first N-dimensional input image (f) is a 3-dimensional image, the second N-dimensional input image (g) is a 3-dimensional image, the N-dimensional form (V g ) is a 3-dimensional image, and the N-dimensional output signal is a 3-dimensional image.

4. The method of claim 1 , further comprising a change identification process including at least one of synthesis of morphological changes or emulation of morphological changes using physics-based algorithms to match or diagnose mechanisms of observed morphological behavior between the second N-dimensional input image (g) and additional N-dimensional input images to determine additional N-dimensional forms for use in the computing.

5. The method of claim 4 , further comprising a learning process including at least one of an adaptive learning capability or a prognostic learning capability that uses the change identification process to predict future trends of surface morphology.

6. The method of claim 5 , wherein the learning process is configured to operate in substantially real-time within a computing system to determine substantially real-time change in images of a surface of an object, the images represented by the second N-dimensional input image (g) and the additional N-dimensional input images; and
further comprising modifying material production of the object responsive to the substantially real-time learning process, wherein substantially real-time is within four seconds.

7. The method of claim 1 , wherein the acts of identifying the N-dimensional form (V g ) computing the average brightness value, and generating the N-dimensional output image signal are performed with a static library of computing algorithms identified for analysis of existing data comprising the second N-dimensional input image (g).

8. The method of claim 1 , wherein the acts of identifying the at least one N-dimensional form (V g ), filtering the N-dimensional input image (g), and generating the N-dimensional output signal is performed with a static library of computing algorithms identified for analysis of existing data and performed at least one additional time with a dynamic library of computing algorithms that adapt to information from additional N-dimensional input images.

9. The method of claim 1 , further comprising increasing a confidence level in the projectional morphological analysis using statistical analyses of results of the projectional morphological analysis from a plurality of N-dimensional signals used as the N-dimensional input images to the projectional morphological analysis.

10. A computing system for performing projectional morphological analysis, comprising:
a memory configured for storing computing instructions; and
a processor operably coupled to the computing system and configured for executing the computing instructions to perform a projection morphological analysis by:
inputting a first N-dimensional input image (f) representative of a scene of a field of view;
defining non-intersecting subsets (A) from the first N-dimensional input image (f) in which points have a constant brightness value (C) within its respective subset (A), a union of the non-intersecting subsets (A) denoted as set X;
inputting second N-dimensional input image (g) representative of the scene of the field of view taken under different conditions, the second N-dimensional input image (g) also defined by the set X having the same defined non-intersecting subsets (A) to denote an N-dimensional form (V g ) of the second N-dimensional input image (g);
determining an N-dimensional form (V g ) of the second N-dimensional input image (g) including a concatenation of the non-intersecting subsets (A) defined by the first N-dimensional input image (f) that have constant brightness values within each respective non-intersecting subset (A);
generating a morphological projection image (g′) by processing the second N-dimensional input image (g) relative to the N-dimensional form (V g ) by taking a difference of a brightness value of each of the points of the N-dimensional signal (g) relative to an average of the brightness value of the N-dimensional form (V g ) within its respective non-intersecting subset (A); and
generating an N-dimensional output signal indicating a morphological difference of the first N-dimensional input image (f) and the second N-dimensional input image (g) by subtracting morphological projection image (g′) from the second N-dimensional input image (g).

11. The computing system of claim 10 , wherein the first N-dimensional input image (f) is a 2-dimensional image, the second N-dimensional input image (g) is a 2-dimensional image, the N-dimensional form (V g ) is a 2-dimensional image, and the N-dimensional output signal is a 2-dimensional image.

12. The computing system of claim 10 , wherein the first N-dimensional input image (f) is a 3-dimensional image, the second N-dimensional input image (g) is a 3-dimensional image, the N-dimensional form (V g ) is a 3-dimensional image, and the N-dimensional output signal is a 3-dimensional image.

13. The computing system of claim 10 , further comprising a change identification software module including computing instructions to perform at least one of synthesis of morphological changes or emulation of morphological changes using physics-based algorithms to match or diagnose mechanisms of observed morphological behavior between the N-dimensional input image (g) and additional N-dimensional input images to determine additional N-dimensional forms for use in the filtering.

14. The computing system of claim 13 , further comprising a learning software module including computing instructions to perform at least one of an adaptive learning capability or a prognostic learning capability that uses the change identification software module to predict future trends of surface morphology.

15. The computing system of claim 14 , wherein the learning software module is configured to operate in substantially real-time to determine substantially real-time change in images of a surface of an object, the images represented by the N-dimensional input image (g) and the additional N-dimensional input

Assignees

• Battelle Energy Alliance Llc

Inventors

• Glazoff Michael V
• Gering Kevin L
• Garnier John E
• Rashkeev Sergey N
• Pyt Ev Yuri Petrovich

Classifications

• G06T2207/10056

  Microscopic image · CPC title

• G06T2207/20036

  Morphological image processing · CPC title

• G06T7/00 Primary

  Image analysis · CPC title

• G06T2207/30136

  Metal · CPC title