Classifying ambiguous image data

Patent metadata

Field	Value
Publication number	US-9747519-B2
Application number	US-201514695623-A
Country	US
Kind code	B2
Filing date	Apr 24, 2015
Priority date	Apr 24, 2015
Publication date	Aug 29, 2017
Grant date	Aug 29, 2017

Abstract

Examples are disclosed herein that relate to classifying ambiguous data in an image. One example provides, on a computing device, a method of processing image data, the method comprising defining a contiguous area of ambiguous data in an image, determining a border surrounding the contiguous area, classifying the contiguous area as dark or far based on a characteristic of one or more pixels of the border, and for the classified contiguous area, processing one or more pixels of the contiguous area differently if the contiguous area is classified as far than if the contiguous area is classified as dark.

First claim

The invention claimed is:

1. On a computing device, a method of processing image data, the method comprising:
defining a contiguous area of ambiguous data in an image;
determining a border surrounding the contiguous area;
classifying the contiguous area as either dark or far based on a characteristic of one or more pixels of the border; and
for the classified contiguous area, processing one or more pixels of the contiguous area differently if the contiguous area is classified as far than if the contiguous area is classified as dark.

2. The method of claim 1 , wherein the image comprises a depth image.

3. The method of claim 1 , further comprising, for each pixel of the border, determining whether the pixel is within a threshold distance of a maximum depth value.

4. The method of claim 3 , wherein classifying the contiguous area as far comprises determining that a ratio of a number of pixels of the border within the threshold distance of the maximum depth value to a total number of pixels of the border exceeds a threshold ratio.

5. The method of claim 3 , wherein classifying the contiguous area as far comprises determining that a ratio of a number of pixels of the border within the threshold distance of the maximum depth value to a total number of pixels of the contiguous area exceeds a threshold ratio.

6. The method of claim 3 , further comprising calculating a weighted total score based on depth values of pixels of the border compared to the maximum depth value, and wherein classifying the contiguous area as far comprises determining that the weighted total score exceeds a threshold score.

7. The method of claim 1 , further comprising determining depth gradients of pixels adjacent to the border, and classifying the contiguous area as far based on the depth gradients determined.

8. The method of claim 1 , wherein if the contiguous area is classified as far, the method comprises removing the contiguous area from a depth map.

9. A computing system, comprising:
a depth camera;
a logic subsystem; and
a storage subsystem comprising instructions executable by the logic subsystem to
define a contiguous area of ambiguous data in an image,
determine a border surrounding the contiguous area,
classify the contiguous area as either dark or far based on a characteristic of one or more pixels of the border, and
for the classified contiguous area, process one or more pixels of the contiguous area differently if the contiguous area is classified as far than if the contiguous area is classified as dark.

10. The computing system of claim 9 , wherein the depth camera comprises a time-of-flight depth camera.

11. The computing system of claim 9 , wherein the computing system comprises a mobile device.

12. The computing system of claim 9 , wherein the instructions are further executable to, for each pixel of the border, determine whether the pixel is within a threshold distance of a maximum depth value.

13. The computing system of claim 12 , wherein the instructions are further executable to classify the contiguous area as far upon determining that a ratio of a number of pixels of the border within the threshold distance of the maximum depth value to a total number of pixels of the border exceeds a threshold ratio.

14. The computing system of claim 12 , wherein the instructions are further executable to classify the contiguous area as far upon determining that a ratio of a number of pixels of the border within the threshold distance of the maximum depth value to a total number of pixels of the contiguous area exceeds a threshold ratio.

15. The computing system of claim 12 , wherein the instructions are further executable to calculate a weighted total score based on depth values of pixels of the border compared to the maximum depth value, and to classify the contiguous area as far upon determining that the weighted total score exceeds a threshold score.

16. The computing system of claim 12 , wherein the instructions are further executable to determine depth gradients of pixels adjacent to the border, and classify the contiguous area as far based on the depth gradients determined.

17. The computing system of claim 9 , wherein if the contiguous area is classified as far, the instructions are further executable to remove the contiguous area from a depth map.

18. A computing system, comprising
a depth camera;
a logic subsystem; and
a storage subsystem comprising instructions executable by the logic subsystem to
define a contiguous area of ambiguous data in an image;
determine a border surrounding the contiguous area;
classify the contiguous area as either dark or far based on a characteristic of one or more pixels of the border;
remove the contiguous area from a depth map if the contiguous area is classified as far; and
include the contiguous area in the depth map if the contiguous area is classified as dark.

19. The computing system of claim 18 , wherein the computing system comprises a mobile device.

20. The computing system of claim 18 , wherein the instructions are further executable to classify the contiguous area as far based on a number of pixels of the border that are within a threshold distance of a maximum depth value.

Assignees

• Microsoft Technology Licensing Llc

Inventors

• Schoenberg Michael John

Classifications

• G06F18/22

  Matching criteria, e.g. proximity measures · CPC title

• G06V20/36 Primary

  Indoor scenes · CPC title

• G06T7/50

  Depth or shape recovery · CPC title

• G06K9/00691

  Physics · mapped topic

• G06K9/46

  Physics · mapped topic