SYSTEMS AND METHODS FOR IDENTIFICATION OF FLUID AND SUBSTRATE COMPOSITION OR PHYSICO-CHEMICAL PROPERTIES

• US 20130073221A1
• Filed: 09/14/2012
• Published: 03/21/2013
• Est. Priority Date: 09/16/2011
• Status: Abandoned Application

• Current Assignee: Columbia University, Iowa State University Research Foundation Incorporated (Iowa State University), Arizona Board of Regents (University of Arizona), University of Arizona
• Original Assignee: Iowa State University Research Foundation Incorporated (Iowa State University)
• Inventors: Attinger, Daniel, Zenhausern, Frederic, Hurth, Cedric, Chang, Shih-Fu, Li, Zhenguo
• Application Number:
  US13/619,338
• Publication Number:
  US 20130073221A1
• Time in Patent Office: Days
• US Class Current: 702/30
• CPC Class Codes: G01N 21/8803 Visual inspection measuring projectors G01B9/08
• CPC Class Codes: G06F 18/24147 Distances to closest patterns, e.g. nearest neighbour classification
• CPC Class Codes: G06V 10/449 Biologically inspired filters, e.g. difference of Gaussians [DoG] or Gabor filters
• CPC Class Codes: G06V 10/467 Encoded features or binary features, e.g. local binary patterns [LBP]
• CPC Class Codes: G06V 10/764 using classification, e.g. of video objects
• CPC Class Codes: G06V 20/695 Preprocessing, e.g. image segmentation
• CPC Class Codes: G06V 20/698 Matching; Classification
• CPC Class Codes: G16C 20/20 Identification of molecular entities, parts thereof or of chemical compositions
• CPC Class Codes: G16C 99/00 Subject matter not provided for in other groups of this subclass

Abstract

Techniques for identifying a composition of a target fluid using a set of vectors representing known residue patterns for a two or more fluids including said target fluid is provided. An exemplary method includes storing one or more digital measurements of residue for the target fluid, extracting one or more descriptive features from the measurements; and processing descriptive features to identify the composition of the target fluid. The processing includes using a machine learning algorithm trained with data linking residue morphology to fluid composition. A distance between a vector representing said one or more descriptive features and said set of vectors representing known residue patterns is determined, and a residue is assigned to one or more of the known residue patterns.

Claims

1. A method for identifying at least one of a physico-chemical property or composition of a target fluid using a set of vectors representing known residue patterns for a two or more fluids including said target fluid, comprising:

acquiring one or more digital measurements of residue for said target fluid;

extracting one or more descriptive features from at least one of said measurements; and

processing one or more of said descriptive features to identify said at least one physico-chemical property or composition of said target fluid from said at least one of said measurements;

wherein said processing includes using a matching algorithm or a machine learning algorithm trained with data linking residue patterns to fluid composition or physico-chemical properties.

2. The method of claim 1, further comprising:

determining a distance between a vector representing said one or more descriptive features and said set of vectors representing known residue patterns, and

assigning a residue to one or more of the known residue patterns that minimizes said distance.

3. The method of claim 1, wherein acquiring one or more digital measurements comprises capturing one or more images.

4. The method of claim 1, wherein said acquiring further comprises:

placing a quantity of a fluid on a substrate, wherein the fluid dries and leaves said residue.

5. The method of claim 1, wherein extracting one or more descriptive features comprises performing automatic localization of the residue as a region of interest in the measurement.

6. The method of claim 5, wherein the one or more measurements comprises one or more images, and automatic localization of the residue comprises:

converting an input image into a binary-formatted image;

determining a largest object in the binary-formatted image to be the residue;

determining a boundary area of the largest object; and

cropping an area in the input image corresponding to the boundary area of the largest object.

7. The method of claim 1, wherein extracting the one or more descriptive features comprises assembling one or more row vectors from discriminative feature vectors having relative weights, the discriminative feature vectors characterizing one or more of a color distribution, a local binary pattern, a Gabor wavelet pattern, and a residue size.

8. The method of claim 7, wherein the discriminative feature vectors characterize a color distribution, the method further comprising computing the color distribution, including:

converting an input image into a color space having a plurality of color channels;

computing a pixel histogram for each of the plurality of color channels;

computing a mean, standard deviation, skew, energy and entropy for each color channel corresponding to a plurality of components for each color channel; and

normalizing each of the plurality of components to a unit vector corresponding to the color distribution.

9. The method of claim 7, wherein the discriminative feature vectors characterize a local binary pattern, the method further comprising computing the binary pattern, including:

labeling each pixel in an input image by thresholding each pixel with a gray level value of a center pixel; and

assigning each pixel with a binary number corresponding to the thresholding of the gray level value.

10. A method for identifying fluid composition data, comprising:

acquiring one or more digital measurements of liquid residue;

extracting one or more descriptive features from at least one of said measurements; and

processing one or more of said descriptive features to classify said at least one measurement;

wherein said processing includes using an unsupervised and trained machine learning technique, and

identifying patterns common to several measurements in a residue dataset, if any, and grouping similar residues into clusters without labeled data, such that each cluster, and each of the similar residues grouped therein, corresponds to a distinct visual pattern.

11. The method of claim 10, wherein said acquiring further comprises:

placing a quantity of a fluid substance on a substrate, wherein the substrate dries and leaves said residue.

12. The method of claim 10, wherein the at least one measurement comprises at least one image, and extracting one or more descriptive features comprises performing automatic localization of the residue as a region of interest in the image.

13. The method of claim 12, wherein automatic localization of the residue comprises:

converting an input image into a binary-formatted image;

determining a largest object in the binary-formatted image to be the residue;

determining a boundary area of the largest object; and

cropping an area in the input image corresponding to the boundary area of the largest object.

14. The method of claim 10, wherein extracting the one or more descriptive features comprises assembling one or more row vectors from discriminative feature vectors having relative weights, the discriminative feature vectors characterizing one or more of a color distribution, a local binary pattern, a Gabor wavelet pattern, and a residue size.

15. The method of claim 14, wherein the one or more measurements comprises at least one image and the discriminative feature vectors characterize a color distribution, the method further comprising computing the color distribution, including:

converting an input image into a color space having a plurality of color channels;

computing a pixel histogram for each of the plurality of color channels;

computing a mean, standard deviation, skew, energy and entropy for each color channel corresponding to five components for each color channel; and

normalizing each of the five components to a unit vector corresponding to the color distribution.

16. The method of claim 14, wherein the discriminative feature vectors characterize a local binary pattern, the method further comprising computing the binary pattern, including:

labeling each pixel in an input image by thresholding each pixel with a gray level value of a center pixel; and

associating each pixel with a sequence of binary numbers corresponding to the gray level value.

17. The method of claim 10, wherein processing one or more of said descriptive features to classify said at least one measurement comprises clustering said at least one measurement and one or more other of the stored digital measurements.

18. A system for identifying at least one of a composition or a property of a target fluid using a set of vectors representing known residue patterns for a two or more fluids including said target fluid, comprising:

one or more memories storing multimedia data;

one or more processors coupled to said one or more memories;

a detection device configured to capture a measurement of residue for said target fluid on a substrate, coupled to said one or more processors and said one or more memories so as to store said measurement in said one or more memories; and

a computer readable medium containing digital information coupled to said one or more processors, said digital information comprising a machine learning algorithm trained with data linking residue morphology to fluid composition or physico-chemical properties, where when executed said digital information causes said one or more processors to;

extract descriptive features from said measurement obtained from said detection device,

determine a distance between a vector representing said one or more descriptive features and said set of vectors representing known residue patterns, and

assign the fluid residue to one or more of the known residue patterns that minimizes said distance.

19. The system of claim 18, wherein the detection device comprises an inverted microscope.

20. The system of claim 19, wherein the microscope comprises one or more objective lenses.

21. The system of claim 18, wherein the detection device comprises a CMOS camera.

22. The system of claim 18, wherein the substrate comprises a solid material

23. The system of claim 18, wherein the substrate comprises a glass slide.

24. The substrate of claim 22, wherein said material is transparent to an electromagnetic wave.

25. The substrate of claim 22, wherein said material is non-transparent to an electromagnetic wave

26. The system of claim 18, wherein the substrate comprises a biomolecular coating.

27. The system of claim 18, wherein the substrate comprises a Streptavidin protein coating.

28. The system of claim 18, further comprising an electromagnetic source configured to illuminate the residue.