Object detection
In an example detection system for vehicles or other objects of interest, objects are detected in real-time at full VGA 30 frame per second resolution. A preprocessor may perform run-length encoding (RLE) to provide detected edges. The image may then be scanned from the bottom up to identify vertical clusters or “stacks” or RLEs. Vertical clusters with low vertical density may be eliminated as poor vehicle candidates. Vertical clusters not eliminated may then be processed with a histogram of gradients algorithm, and confirmed with a support vector machine algorithm. A range to the nearest object may also be calculated, and a warning provided if the object is too close.
1. A method of detecting an object of interest in an image comprising:
receiving a first frame of the image;
performing edge detection on the first frame of the image;
performing run-length encoding (RLE) on detected edges of the first frame of the image to identify edges for a shape bounding a candidate object;
generating a hypothesis that the candidate object is the object of interest comprising identifying vertical RLE clusters as part of the candidate object, and processing the vertical RLE clusters upward from a bottom by an intensity contrast; and
confirming in a second frame of the image the hypothesis that the candidate object is the object of interest.
2. The method of claim 1 , wherein generating a hypothesis comprises identifying vertical RLE clusters as vehicle candidates.
3. The method of claim 2 , wherein generating a hypothesis further comprises eliminating vertical RLE clusters with low vertical density.
4. The method of claim 1 , further comprising:
calculating a distance to the object of interest; and
providing a warning if the object of interest is closer than a threshold.
5. The method of claim 1 , wherein confirming the hypothesis comprises:
calculating a histogram of gradients for the candidate object; and
classifying the candidate object as the object of interest using a support vector machine method.
6. The method of claim 1 , further comprising comparing the object of interest to a substantially similar object of interest identified in a next frame to determine temporal continuity.
7. A tangible computer-readable storage medium having stored thereon executable instructions operable to:
receive a first frame of an image;
perform edge detection on the first frame of the image;
perform run-length encoding (RLE) on detected edges to identify edges for a shape bounding a candidate object;
generate a hypothesis that the candidate object is an object of interest comprising identifying vertical RLE clusters as part of the candidate object, and processing the vertical RLE clusters upward from a bottom by an intensity contrast; and
confirm in a second frame of the image the hypothesis that the candidate object is the object of interest.
8. The medium of claim 7 , wherein the instructions operable to generate a hypothesis comprise instructions operable to identify vertical RLE clusters as vehicle candidates.
9. The medium of claim 8 , wherein the instructions operable to generate a hypothesis further comprises instructions operable to eliminate vertical RLE clusters with low vertical density.
10. The medium of claim 7 , further comprising instructions operable to:
calculate a distance to the object of interest; and
provide a warning if the object of interest is closer than a threshold.
11. The medium of claim 7 , wherein the instructions operable to confirm the hypothesis comprise instructions operable to:
calculate a histogram of gradients for the candidate object; and
classify the candidate object as the object of interest using a support vector machine method.
12. The medium of claim 7 , further comprising instructions operable to compare the object of interest to a substantially similar object of interest identified in a next frame to determine temporal continuity.
13. An advanced driver assistance system (ADAS) comprising:
a camera;
a processor; and
a storage medium having stored thereon executable instructions operable to instruct the processor to:
receive a first frame of an image from the camera;
perform edge detection on the first frame of the image;
perform run-length encoding (RLE) on detected edges to identify edges for a shape bounding a candidate object;
generate a hypothesis that the candidate object is an object of interest comprising identifying vertical RLE clusters as part of the candidate object, and processing the vertical RLE clusters upward from a bottom by an intensity contrast; and
confirm in a second frame of an image the hypothesis that the candidate object is the object of interest.
14. The ADAS of claim 13 , wherein the instructions operable to generate a hypothesis comprise instructions operable to identify vertical RLE clusters as vehicle candidates.
15. The ADAS of claim 14 , wherein the instructions operable to generate a hypothesis further comprises instructions operable to eliminate vertical RLE clusters with low vertical density.
16. The ADAS of claim 13 , further comprising instructions operable to:
calculate a distance to the object of interest; and
provide a warning if the object of interest is closer than a threshold.
17. The ADAS of claim 13 , wherein the instructions operable to confirm the hypothesis comprise instructions operable to:
calculate a histogram of gradients for the candidate object; and
classify the candidate object as the object of interest using a support vector machine method.
18. The ADAS of claim 13 , further comprising instructions operable to compare the object of interest to a substantially similar object of interest identified in a next frame to determine temporal continuity.