Optical images by quadrupole convolution

• US 5,027,419 A
• Filed: 03/31/1989
• Issued: 06/25/1991
• Est. Priority Date: 03/31/1989
• Status: Expired due to Fees

• Current Assignee: Atomic Energy of Canada Limited (Government of Canada)
• Original Assignee: Atomic Energy of Canada Limited (Government of Canada)
• Inventors: Davis, Ronald S.
• Primary Examiner(s): Boudreau, Leo H.
• Assistant Examiner(s): KLOCINSKI, STEVEN
• Application Number:
  US07/330,895
• Time in Patent Office: 816 Days
• Field of Search: 382/28, 382/42, 382/27, 382/65
• US Class Current: 382/206
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/18019 by matching or filtering

Abstract

The invention described is a process of and an apparatus for recognizing the size, location, orientation etc. of an object without human intervention. The convolution technique is used to solve the field theory equations to generate linearity signals characteristic of points in the optical image of the object. The linearity signal is perception of linear or strip-like features in an image. The recognition is achieved by analyzing the linearity signals.

Claims

1. A machine recognition process for recognizing the shape, location or orientation of an object by determining the linearity in an optical image of the said object in an (x,y) Cartesian coordinate system, comprising steps of:

capturing the said optical image in an image field defined in the said (x,y) system,

generating an image signal v(x'"'"',y'"'"') for each point (x'"'"',y'"'"') in the said image field,

deriving a linearity signal L(x,y) for each point (x,y) in the said image field by means of convolution expressed by the following equation;

##EQU8## where Q(x-x'"'"',y-y'"'"') is the kernel and has the value;

space="preserve" listing-type="equation">Q(x-x'"'"',y-y'"'"')=1/ ((x-x'"'"')+i(y-y'"'"')).sup.2

3. The machine recognition process according to claim 1, wherein:

the said step of capturing the said optical image includes a step of scanning the said image field in a raster pattern so that the said image field is made up of a matrix of discrete pixels defined in the said (x,y) system, and

the said step of deriving the linearity signal L(x,y) further comprises a step of approximating the convolution by summation with respect to the said discrete pixels by the following equation;

##EQU9##

2. where i is the square root of -1,

selecting a certain set of said linearity signals for points in the said image field, and

analyzing the said set of linearity signals in view of prestored reference parameters so that the optical image of the object is recognized.

4. The machine recognition process according to claim 2, wherein:

the said step of deriving the linearity signal L(x,y) further comprises a step of converting parameters between the said cartesian coordinate system and a polar coordinate system, using the following equations;

space="preserve" listing-type="equation">x=r cos θ

space="preserve" listing-type="equation">y=r sin θ

6. The machine recognition process according to claim 2, wherein:

the said step of selecting a certain set of said linearity signals is being performed as the image field is being scanned.

7. The machine recognition process according to claim 2, wherein:

the said step of selecting a certain set of said linearity signals is being performed according to a predetermined region-of-interest criterion.

8. The machine recognition process according to claim 2, wherein:

the said step of deriving a linearity signal L(x,y) for each point (x,y) in the said image field comprises;

generating at each pixel a physical process which obeys Laplace'"'"'s equation in two dimensions expressed by the following equation;

##EQU10## where Q(x-x'"'"',y-y'"'"') is the kernel.

9. The machine recognition process according to claim 6, wherein:

the said physical process is the conduction of electricity in a resistive sheet.

10. The machine recognition process according to claim 6, wherein:

the said physical process is the conduction of heat in a heat conductive sheet.

11. The machine recognition process according to claim 6, wherein:

the said physical process is the deformation of a membrane.

14. The image recognition apparatus for recognizing an object according to claim 10, wherein:

the said image means for optically capturing the said optical image includes scanning means for scanning the said image field so that the said image field is made up of a matrix of discrete pixels defined in the said (x,y) system, and

the said linearity signal means derives the linearity signal L(x,y) for each pixel located at a point (x,y) in the said image field and further sums the said linearity signals L(x,y) for all the pixels.

15. The image recognition apparatus according to claim 11, wherein:

the said linearity signal means for deriving the linearity signal comprises;

Laplace means for generating at each pixel a physical process which obeys Laplace'"'"'s equation in two dimensions expressed by the following equation;

##EQU12## where Q(x-x'"'"',y-y'"'"') is the kernel.

5. wherein r and θ

are polar coordinate values.

12. An image recognition apparatus for recognizing the shape, location or orientation of an object by determining the linearity in an optical image of the said object in an (x,y) cartesian coordinate system, comprising:

image means for optically capturing the said optical image in an image field defined in the said (x,y) system,

image signal means for generating an image signal v(x'"'"',y'"'"') for each point (x'"'"',y'"'"') in the said image field,

linearity signal means for deriving a linearity signal L(x,y) for each point (x,y) in the said image field by means of convolution expressed by the following equation;

##EQU11## where Q(x-x'"'"',y-y'"'"') is the kernel and has the value;

space="preserve" listing-type="equation">Q(x-x'"'"',y-y'"'"')=1/ ((x-x'"'"')+i(y-y'"'"')).sup.2

16. The image recognition apparatus according to claim 12, wherein:

the said physical process is the conduction of electricity in a resistive sheet.

17. The image recognition apparatus according to claim 12, wherein:

the said physical process is the conduction of heat in a heat conductive sheet.

18. The image recognition apparatus according to claim 12, wherein:

the said physical process is the deformation of a membrane.

13. where i is the square root of -1,

selection means for selecting a certain set of said linearity signals for points in the said image field, and

analysis means for analyzing the said set of linearity signals in view of prestored reference parameters so that the optical image of the object is recognized.