OBJECT RECOGNITION SYSTEM IMPLEMENTING IMAGE DATA TRANSFORMATION

• US 20140205139A1
• Filed: 01/18/2013
• Published: 07/24/2014
• Est. Priority Date: 01/18/2013
• Status: Abandoned Application

• Current Assignee: Caterpillar Incorporated
• Original Assignee: Caterpillar Incorporated
• Inventors: KRIEL, Bradley Scott, Morris, Daniel
• Application Number:
  US13/745,637
• Publication Number:
  US 20140205139A1
• Time in Patent Office: Days
• US Class Current: 382/103
• CPC Class Codes: G06V 20/58 Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads

Abstract

A object recognition system has a camera configured to generate source image data and a processor configured to access the source image data from the camera. The processor is also configured to accesses state data of the camera and generate transformed image data from the source image data based at least in part on the state data. The processor is also configured to detect an object in the transformed image data and to classify the detected object using the transformed image data.

Claims

1. An object recognition system comprising:

a camera configured to generate source image data; and

a processor configured to;

access the source image data from the camera;

access state data of the camera;

generate transformed image data from the source image data based at least in part on the state data;

detect an object in the transformed image data; and

classify the detected object using the transformed image data.

2. The system of claim 1, wherein the state data includes a down-vector value correlating with a pixel location of the source image data.

3. The system of claim 2 wherein the processor is further configured to:

access a minimum radius value corresponding to a first radius of pixels from the down-vector value;

access a maximum radius value corresponding to a second radius of pixels from the down-vector value; and

,

generate the transformed image data by using a portion of the source image data corresponding to pixels that are between the minimum radius value and the maximum radius value.

4. The system of claim 3 wherein the detected object is within the minimum radius and the maximum radius.

5. The system of claim 1, wherein the processor is configured to generate the transformed image data by mapping pixels of the image source data to a pixel map corresponding to the transformed image data.

6. The system of claim 1 wherein:

the camera is mounted to a machine; and

,

the processor is further configured to render the source image data on a display mounted to the machine.

7. The system of claim 6 wherein the display image includes an indication of the detected object.

8. The system of claim 7 wherein the indication includes a boundary box that is colored based at least in part on the type of the detected object.

9. The system of claim 7 wherein the indication includes text describing the type of the detected object.

10. The system of claim 6 wherein the display image includes a first indication when the detected object is of a first type and a second indication when the detected object is of a second type.

11. A method for recognizing objects comprising:

accessing source image data from a camera;

accessing state data of the camera;

generating transformed image data using the source image data, the generating being based at least in part on the state data;

detecting an object in the transformed image data; and

classifying the detected object using the transformed image data.

12. The method of claim 11, wherein the state data includes a down-vector value correlating with a pixel location of the source image data.

13. The method of claim 12 further including:

accessing a minimum radius value corresponding to a first radius of pixels from the down-vector value;

accessing a maximum radius value corresponding to a second radius of pixels from the down-vector value; and

,

generating the transformed image data by using a portion of the source image data corresponding to pixels that are between the minimum radius value and the maximum radius value.

14. The method of claim 11, wherein the transformed image data is generated by mapping pixels of the source image data to a pixel map corresponding to the transformed image data.

15. The method of claim 11, wherein the camera is mounted to a machine and the method further includes rendering the source image data on a display mounted to the machine.

16. The method of claim 15 wherein the display image includes an indication of the detected object.

17. The method of claim 16 wherein the indication includes a boundary box that is colored based at least in part on the type of the detected object.

18. The method of claim 16 wherein the indication includes text describing the type of the detected object.

19. The method of claim 15 wherein the display image includes a first indication when the detected object of interest is of a first type and a second indication when the detected object is of a second type.

20. A mobile machine comprising:

a cabin;

a display disposed within the cabin;

a frame;

a camera connected to the frame configured to generate image data;

a processor in communication with the camera and the display, the processor configured to;

access source image data from the camera;

access state data of the camera, the state data including;

a center pixel location corresponding to the down vector of the camera,

a minimum radius value, and

a maximum radius value.

generate transformed image data using the source image by using a portion of the source image data between the minimum radius value and the maximum radius value;

identify an object of interest in the transformed image data;

classify the identified object of interest using the transformed image data;

render the source image data on the display, wherein the rendering includes an indication of the location of the object of interest and the classification of the object of interest.