Method and device for processing graphics data in graphics processing unit

Patent metadata

Field	Value
Publication number	US-9959661-B2
Application number	US-201615367507-A
Country	US
Kind code	B2
Filing date	Dec 2, 2016
Priority date	Dec 2, 2015
Publication date	May 1, 2018
Grant date	May 1, 2018

Abstract

Provide are a methods and devices for processing graphics data in a graphics processing unit (GPU). The method of processing graphics data includes receiving, at a processor, a difference of Gaussian (DOG) layer of an image, detecting, from the received DOG layer, a candidate DOG layer of the image as an intermediate layer, detecting at least one extreme point by comparing values of the candidate DOG layer with values of a previous DOG layer and a next DOG layer, and storing the at least one extreme point in a buffer.

First claim

What is claimed is:

1. A method of processing graphics data by executing multiple pipeline passes of a processor to detect a feature of an image, the method comprising:
receiving, at a processor, a difference of Gaussian (DOG) layer of the image, the DOG layer generated by pixel subtraction of adjacent scale space layers of the image from each other;
detecting, from the received DOG layer, a candidate DOG layer of the image as an intermediate layer;
detecting at least one extreme point of the image by comparing corresponding pixel values of the candidate DOG layer with pixel values of a previous DOG layer and a next DOG layer; and
storing the at least one extreme point of the feature of the image in a buffer; and
wherein the multiple pipeline passes of the processor include a first pass for generating the DOG layer of the image, a second pipeline pass for extrema detection, and a third pipeline pass for performing key point localization.

2. The method of claim 1 , wherein key point localization is performed at a shader core of the processor using the stored at least one extreme point detected during the second pipeline pass.

3. The method of claim 1 , wherein the detecting of the at least one extreme point comprises:
comparing the pixel values of the candidate DOG layer with the corresponding pixel values of the previous DOG layer to detect extreme points; and
comparing the extreme points of the candidate DOG layer and the previous DOG layer with the corresponding pixel values of the next DOG layer to detect the at least one extreme point.

4. The method of claim 1 , wherein values of the DOG layer are stored in the buffer.

5. The method of claim 1 , wherein the DOG layer is computed via a shader core of the processor based on tiles of the image comprised of pixels.

6. The method of claim 1 , wherein the at least one extreme point comprises any one or any combination of a maximum value and a minimum value.

7. A non-transitory computer-readable recording medium storing instructions that, when executed by a processor, cause the processor to perform the method of claim 1 .

8. A device for processing data of extreme points, the device comprising:
a graphics processing unit (GPU) configured to:
receive a candidate difference of Gaussian (DOG) layer of an image, the DOG layer generated by pixel subtraction of adjacent scale space layers of the image from each other,
detect the candidate DOG layer of the image as an intermediate layer, and
detect at least one extreme point by comparing corresponding pixel values of the candidate DOG layer with values of a previous DOG layer and a next DOG layer; and
a buffer configured to store the at least one extreme point,

wherein the GPU is configured to perform multiple pipeline passes to detect the at least one extreme point.

9. The device of claim 8 , further comprising:
a shader core configured to receive the stored at least one extreme point, and to detect key point localization of the received at least one extreme point.

10. The device of claim 8 , wherein the GPU is further configured to:
compare the values of the candidate DOG layer with the corresponding pixel values of the previous DOG layer to detect extreme points, and
compare the extreme points of the candidate DOG layer and the previous DOG layer with the corresponding values of the next DOG layer to detect the at least one extreme point.

11. The device of claim 8 , wherein values of the DOG layer are stored in the buffer.

12. The device of claim 11 , further comprising a shader core configured to compute the DOG layer based on tiles of the image comprised of pixels.

13. The device of claim 8 , wherein the at least one extreme point comprises any one or any combination of a maximum value and a minimum value.

14. A graphics processing unit (GPU) to detect at least one extreme point in an image, the GPU comprising:
a shader core configured to determine a difference of Gaussian (DOG) layer of an image, the DOG layer generated by pixel subtraction of adjacent scale space layers of the image from each other;
a comparator configured to identify the DOG layer as an intermediate layer, and to detect the at least one extreme point by comparing corresponding pixel values of the DOG layer with values of a previous DOG layer and a next DOG layer; and
a buffer configured to store values of the DOG layer, the previous layer, the next layer and the extreme point,
wherein the GPU is configured to perform multiple pipeline passes to detect the at least one extreme point.

15. The GPU of claim 14 , wherein the comparator is further configured to:
obtain pixel values of the DOG layer from the buffer;
compare the pixel values of the DOG layer with the values of the previous layer, stored in the buffer, to detect first extreme points;
compare the first extreme points with the values of the next layer, stored in the buffer, to detect the at least one extreme point.

16. The GPU of claim 14 , wherein the shader core is further configured to receive the detected at least one extreme point, during one of the multiple passes of the pipeline to perform key point localization on the received at least one extreme point.

Assignees

• Samsung Electronics Co Ltd

Inventors

• Deshwal Ankur
• Singhal Somya
• Varadarajan Keshavan
• Kohli Soma

Classifications

• G06T15/005 Primary

  General purpose rendering architectures · CPC title

• G06T11/40

  Filling planar surfaces by adding surface attributes, e.g. adding colours or textures · CPC title

• G06T1/20

  Processor architectures; Processor configuration, e.g. pipelining · CPC title