Image processing methods and arrangements useful in automated store shelf inspections

1. A method of checking inventory in a retail store, the store including products stocked on shelves positioned adjacent an aisle, the method including the acts:

producing shelf imagery using a robot that navigates along said aisle of said retail store, the robot being equipped with an imaging system comprising one or more cameras and at least one illumination source;

processing a low-resolution counterpart to said shelf imagery to sense presence of machine-readable information at a particular location within the shelf imagery;

based on said particular location, identifying a region of interest containing the machine-readable information; and

analyzing a high-resolution counterpart to said region of interest, depicted in said shelf imagery, to decode an identifier from said machine-readable information;

wherein processing of the shelf imagery is reduced due to analysis of said region of interest, rather than all said shelf imagery, to decode said identifier;

the method further including detecting, from said low-resolution counterpart to said shelf imagery, spatial frequency impulses associated with an orientation signal of a digital watermark, said spatial frequency impulses indicating presence of machine-readable digital watermark information.

2. The method of claim 1 that includes stitching together plural image frames captured by said imaging system to produce panoramic imagery, said panoramic imagery serving as said shelf imagery.

3. A method of checking inventory in a retail store, the store including products stocked on shelves positioned adjacent an aisle, the method including the acts:

producing shelf imagery using a robot that navigates along said aisle of said retail store, the robot being equipped with an imaging system comprising one or more cameras and at least one illumination source;

processing a low-resolution counterpart to said shelf imagery to sense presence of machine-readable information at a particular location within the shelf imagery;

based on said particular location, identifying a region of interest containing the machine-readable information; and

analyzing a high-resolution counterpart to said region of interest, depicted in said shelf imagery, to decode an identifier from said machine-readable information;

wherein processing of the shelf imagery is reduced due to analysis of said region of interest, rather than all said shelf imagery, to decode said identifier; and

the machine-readable information is sensed from an electronic shelf label.

4. The method of claim 3 that further includes the robot emitting a signal causing the electronic shelf label to change its display.

5. A method of checking inventory in a retail store, the store including products stocked on shelves positioned adjacent an aisle, the method including the acts:

producing shelf imagery using a robot that navigates along said aisle of said retail store, the robot being equipped with an imaging system comprising one or more cameras and at least one illumination source;

processing a low-resolution counterpart to said shelf imagery to sense presence of machine-readable information at a particular location within the shelf imagery;

based on said particular location, identifying a region of interest containing the machine-readable information; and

analyzing a high-resolution counterpart to said region of interest, depicted in said shelf imagery, to decode an identifier from said machine-readable information;

wherein processing of the shelf imagery is reduced due to analysis of said region of interest, rather than all said shelf imagery, to decode said identifier;

the method further including varying illumination emitted by said illumination source as the robot navigates along the aisle.

6. The method of claim 5 in which said varying is triggered by image data collected by said imaging system.

7. A method of checking inventory in a retail store, the store including products stocked on shelves positioned adjacent an aisle, the method including the acts:

producing shelf imagery using a robot that navigates along said aisle of said retail store, the robot being equipped with an imaging system comprising one or more cameras and at least one illumination source;

processing a low-resolution counterpart to said shelf imagery to sense presence of machine-readable information at a particular location within the shelf imagery;

based on said particular location, identifying a region of interest containing the machine-readable information; and

analyzing a high-resolution counterpart to said region of interest, depicted in said shelf imagery, to decode an identifier from said machine-readable information;

wherein processing of the shelf imagery is reduced due to analysis of said region of interest, rather than all said shelf imagery, to decode said identifier;

the method further including varying a spectrum of illumination emitted by said illumination source as the robot navigates along the aisle.

8. A method of checking inventory in a retail store, the store including products stocked on shelves positioned adjacent an aisle, the method including the acts:

producing shelf imagery using a robot that navigates along said aisle of said retail store, the robot being equipped with an imaging system comprising one or more cameras and at least one illumination source;

processing a low-resolution counterpart to said shelf imagery to sense presence of machine-readable information at a particular location within the shelf imagery;

based on said particular location, identifying a region of interest containing the machine-readable information; and

analyzing a high-resolution counterpart to said region of interest, depicted in said shelf imagery, to decode an identifier from said machine-readable information;

wherein processing of the shelf imagery is reduced due to analysis of said region of interest, rather than all said shelf imagery, to decode said identifier; and

said decoded identifier indicates a first product that should be stocked in an adjoining product region, the method further including:

consulting a data structure to determine a width of said product region for the indicated first product;

by reference to information including a position of the detected shelf label in the shelf imagery, and said determined width of the product region, identifying an excerpt of the shelf imagery within which the first product should be depicted; and

identifying one or more instances of product items depicted within said identified excerpt.

9. The method of claim 8 that includes identifying an instance of a product item within said identified excerpt that is different than said first product, and reporting same as an exception.

10. The method of claim 8 in which the machine-readable information is presented on a medium having a known width, and the method includes using said known width as a measuring reference in identifying said excerpt of the shelf imagery within which the first product should be depicted.

11. The method of claim 10 that includes determining, from the shelf imagery, a number of pixels per inch for depiction of said medium of known width, and using said determined number to establish a width of said excerpt, in pixels.

12. The method of claim 10 that includes determining, from the shelf imagery, a pixel-width of said medium, and multiplying said pixel-width by a ratio between the width of the product region and the width of the medium to determine a pixel-width of the product region.

13. The method of claim 10 that includes using a known width of the medium to limit a search space for recognizing instances of product items depicted within said identified excerpt, since a range of possible scales at which said items may be depicted within said excerpt can thereby be estimated.

14. The method of claim 10 that includes identifying fewer instances of the first product within said product region than expected, and reporting same as an exception.

15. The method of claim 10 in which the camera system includes one camera oriented to view towards the horizon, and a second camera oriented to view above the horizon, wherein the shelf imagery is produced using image data collected by the second camera, and the method includes processing said image data collected by the second camera to compensate for perspective distortion resulting from said view above the horizon.

16. The method of claim 1 in which the machine-readable information is sensed from an electronic shelf label.

17. The method of claim 1 that includes varying illumination emitted by said illumination source as the robot navigates along the aisle.

18. The method of claim 1 that includes varying a spectrum of illumination emitted by said illumination source as the robot navigates along the aisle.

19. The method of claim 1 in which said decoded identifier indicates a first product that should be stocked in an adjoining product region, and the method further includes:

consulting a data structure to determine a width of said product region for the indicated first product;

by reference to information including a position of the detected shelf label in the shelf imagery, and said determined width of the product region, identifying an excerpt of the shelf imagery within which the first product should be depicted; and

identifying one or more instances of product items depicted within said identified excerpt.

20. The method of claim 3 that further includes varying illumination emitted by said illumination source as the robot navigates along the aisle.

21. The method of claim 3 that further includes varying a spectrum of illumination emitted by said illumination source as the robot navigates along the aisle.

22. The method of claim 3 in which said decoded identifier indicates a first product that should be stocked in an adjoining product region, and the method further includes:

consulting a data structure to determine a width of said product region for the indicated first product;

by reference to information including a position of the detected shelf label in the shelf imagery, and said determined width of the product region, identifying an excerpt of the shelf imagery within which the first product should be depicted; and

identifying one or more instances of product items depicted within said identified excerpt.

23. The method of claim 5 that further includes varying a spectrum of illumination emitted by said illumination source as the robot navigates along the aisle.

24. The method of claim 5 in which said decoded identifier indicates a first product that should be stocked in an adjoining product region, and the method further includes:

consulting a data structure to determine a width of said product region for the indicated first product;

by reference to information including a position of the detected shelf label in the shelf imagery, and said determined width of the product region, identifying an excerpt of the shelf imagery within which the first product should be depicted; and

identifying one or more instances of product items depicted within said identified excerpt.

25. The method of claim 7 in which said decoded identifier indicates a first product that should be stocked in an adjoining product region, and the method further includes:

consulting a data structure to determine a width of said product region for the indicated first product;

by reference to information including a position of the detected shelf label in the shelf imagery, and said determined width of the product region, identifying an excerpt of the shelf imagery within which the first product should be depicted; and

identifying one or more instances of product items depicted within said identified excerpt.