DETECTION OF OBJECTS IN IMAGES

• US 20100246890A1
• Filed: 03/26/2009
• Published: 09/30/2010
• Est. Priority Date: 03/26/2009
• Status: Abandoned Application

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Muktinutalapati, Kartik, Tabb, Mark, Ofek, Eyal, Omer, Ido, Kroepfl, Michael
• Application Number:
  US12/411,398
• Publication Number:
  US 20100246890A1
• Time in Patent Office: Days
• US Class Current: 382/105
• CPC Class Codes: G06V 20/63 Scene text, e.g. street names

Abstract

A system described herein includes a detector component that automatically determines location of a license plate in a digital image. The system further includes a blurrer component that automatically blurs the digital image at the determined location of the license plate, wherein blurring undertaken by the blurrer component is based at least in part upon confidence scores assigned to pixels in the digital image that correspond to the determined location of the license plate.

Claims

1. A method comprising the following computer-executable acts:

receiving a digital image that comprises a plurality of pixels;

determining a first gradient value for at least one pixel in the plurality of pixels; and

labeling the at least one pixel as being a particular object based at least in part upon the determined first gradient value for the at least one pixel.

2. The method of claim 1, further comprising:

determining a second gradient value for the at least one pixel in the plurality of pixels; and

labeling the at least one pixel as being the particular object based at least in part upon the determined second gradient value for the at least one pixel.

3. The method of claim 2, wherein the first gradient value is a horizontal gradient value and the second gradient value is a vertical gradient value.

4. The method of claim 3, wherein the at least one pixel is labeled as being a portion of a license plate of a vehicle.

5. The method of claim 4, wherein the first gradient value and the second gradient value are one of intensity gradient values or color gradient values.

6. The method of claim 5, further comprising:

defining a window of pixels that includes a plurality of pixels, wherein the at least one pixel is included in the window of pixels;

determining an average intensity value for the plurality of pixels in the window of pixels;

determining a median intensity value for the plurality of pixels in the window of pixels; and

labeling the at least one pixel as being the particular object based at least in part upon the determined average intensity value and the determined median intensity value.

7. The method of claim 1, further comprising:

determining gradient values for a plurality of pixels in the digital image; and

labeling a subset of the plurality of pixels in the digital image as being at least a portion of the object.

8. The method of claim 7, further comprising visually altering the subset of the plurality of pixels in the digital image to cause at least a portion of the digital image to be blurred.

9. The method of claim 8, further comprising:

analyzing distance data representative of distance between a camera that captured the digital image and the object; and

visually altering the subset of the plurality of pixels based at least in part upon the analyzed distance.

10. The method of claim 1, wherein the gradient is a rate of change of intensity pertaining to the pixel with respect to at least one other pixel.

11. The method of claim 1, wherein the object is a number on a house or a sign on a building facade.

12. The method of claim 1, further comprising:

defining a portion of the digital image that possibly includes the object; and

executing scans horizontally and vertically only in the portion of the digital image that possibly includes the object to determine horizontal and vertical gradient values for each pixel in the portion of the digital image that possibly includes the object.

13. The method of claim 1, further comprising:

analyzing distance data between a camera that captured the digital image and the object; and

labeling the pixel as the object based at least in part upon the analyzed distance.

14. The method of claim 1, further comprising:

analyzing the digital image to determine one or more planes in the digital image; and

labeling the pixel as being at least a portion of the object based at least in part upon the one or more determined planes.

15. A system comprising the following computer-executable components:

a detector component that automatically determines location of a license plate in a digital image; and

a blurrer component that automatically blurs the digital image at the determined location of the license plate, wherein blurring undertaken by the blurrer component is based at least in part upon confidence scores assigned to pixels in the digital image that correspond to the determined location of the license plate.

16. The system of claim 15, wherein the detector component comprises:

a gradient determiner component that determines a horizontal gradient value and a vertical gradient value for at least one pixel in the digital image; and

a labeler component that automatically labels the at least one pixel in the digital image as being the license plate of the vehicle based at least in part upon the determined horizontal gradient value and the determined vertical gradient value.

17. The system of claim 16, wherein the confidence scores are based at least in part upon depth data corresponding to the pixels.

18. The system of claim 16, wherein the horizontal gradient value represents a rate of change in intensity of the at least one pixel with respect to at least one other pixel located horizontally from the at least one pixel and the vertical gradient value represents a rate of change of intensity of the at least one pixel with respect to at least one other pixel located vertically from the at least one pixel.

19. The system of claim 18, wherein the detector component further comprises:

a definer component that defines a window of pixels that includes the at least one pixel;

an average determiner component that determines an average intensity value of pixels in the window of pixels; and

a median determiner component that determines a median intensity value of pixels in the window of pixels, wherein the labeler component labels the at least one pixel as being a portion of the license plate based at least in part upon the determined average intensity value and the determined median intensity value.

20. A computer-readable medium comprising instructions that, when executed by a processor, perform the following acts:

receiving a digital image, wherein the digital image comprises an image of a license plate;

determining horizontal gradient values for a plurality of pixels in the digital image, wherein a horizontal gradient value is representative of a rate of change of intensity of a pixel with respect to at least one other pixel in a horizontal direction in the digital image;

determining vertical gradient values for the plurality of pixels in the digital image, wherein a vertical gradient value is representative of a rate of change of intensity of a pixel with respect to at least one other pixel in a vertical direction in the digital image;

defining a pixel window that comprises multiple pixels;

determining an average intensity value of the multiple pixels in the pixel window;

determining a median intensity value of the multiple pixels in the pixel window;

for at least one pixel in the digital image, determining whether the at least one pixel is included in the image of the license plate based at least in part upon the horizontal gradient value for the at least one pixel, the vertical gradient value for the at least one pixel, the average intensity value of the multiple pixels in the pixel window, wherein the pixel window includes the at least one pixel, and the median intensity value for the plurality of pixels in the pixel window that includes the at least one pixel; and

labeling the pixel as being included in the image of the license plate.