Implementation of Hough transform and its application in video motion analysis
View Patent ↗An implementation of Hough transform is described. According to one aspect, the Hough transform is used to perform motion analysis of patterns in a video using one-dimensional detectors and a one-dimensional voting array.
1. A method of detecting one or more motion models in video frames comprising:
selecting a one-dimensional detector from data items in a data item pool for a video frame;
mapping the detector onto a one-dimensional voting array in a corresponding parameter space;
locating points of projection of data items on the detector; and
establishing a motion model corresponding to a location in the parameter space where the data item projections are dense.
2. The method of claim 1 , further comprising:
removing, from the data item pool, the data items that fit the motion model.
3. The method of claim 1 , further comprising:
selecting the detector from three randomly chosen non-parallel straight lines.
4. The method of claim 1 , wherein each projection is represented by a cell in the voting array.
5. The method of claim 4 , further comprising:
identifying a straight line as a subsequent detector whose projection is substantially close to the location on the detector where the projections is dense and is substantially perpendicular to the detector.
6. The method of claim 4 , wherein the establishing the motion model includes establishing the motion model where the data item projections substantially converge in the parameter space with a cell in the voting array associated with a global maximum.
7. A machine-readable medium having instructions to cause a machine to perform a method of detecting one or more motion models in video frames, the method comprising:
selecting a one-dimensional detector from data items in a data item pool for a video frame;
mapping the detector onto a one-dimensional voting array in a corresponding parameter space;
location points of projection of data items on the detector; and
establishing a motion model corresponding to a location in the parameter space where the data item projections are dense.
8. The machine-readable medium of claim 7 , further comprising:
removing, from the data item pool, the data items that fit the motion model.
9. The machine-readable medium of claim 7 , further comprising:
selecting the detector from three randomly chosen non-parallel straight lines.
10. The machine-readable medium of claim 7 , wherein each projection is represented by a cell in the voting array.
11. The machine-readable medium of claim 10 , further comprising:
identifying a straight line as a subsequent detector whose projection is substantially close to the location on the detector where the projections is dense and is substantially perpendicular to the detector.
12. The machine-readable medium of claim 10 , wherein the establishing the motion model includes establishing the motion model where the data item projections substantially converge in the parameter space with a cell in the voting array associated with a global maximum.
13. A system comprising:
a processor coupled to a memory through a bus; and
a detection process executed from the memory by the processor to cause the processor to select a one-dimensional detector from data items in a data item pool for a video frame, map the detector onto a one-dimensional voting array in a corresponding parameter space, locate points of projection of data items on the detector, establish a motion model corresponding to a location in the parameter space where the data item projections are dense.
14. The system of claim 13 , wherein the detection process further causes the processor to remove, from the data item pool, the data items that fit the motion model.
15. The system of claim 13 , wherein the detection process further causes the processor to select the detector from three randomly chosen non-parallel straight lines.
16. The system of claim 13 , wherein each projection is represented by a cell in the voting array.
17. The system of claim 16 , wherein the detection process further causes the processor to identify a straight line as a subsequent detector whose projection is substantially close to the location on the detector where the projections is dense and is substantially perpendicular to the detector.
18. The system of claim 16 , wherein the detection process further causes the processor, when establishing the motion model, to establish the motion model where the data item projections substantially converge in the parameter space with a cell in the voting array associated with a global maximum.
19. A system comprising:
a means for processing coupled to a means for storing instructions for the means for processing through a means for communicating; and
a detection process executed from the means for storing by the means for processing to cause the means for processing to select a one-dimensional detector from data items in a data item pool for a video frame, map the detector onto a one-dimensional voting array in a corresponding parameter space, locate points of projection of data items on the detector, establish a motion model corresponding to a location in the parameter space where the data item projections are dense.
20. The system of claim 19 , wherein the detection process further causes the means for processing to remove, from the data item pool, the data items that fit the motion model.
21. The system of claim 19 , wherein the detection process further causes the means for processing to select the detector from three randomly chosen non-parallel straight lines.
22. The system of claim 19 , wherein each projection is represented by a cell in the voting array.
23. The system of claim 22 , wherein the detection process further causes the means for processing to identify a straight line as a subsequent detector whose projection is substantially close to the location on the detector where the projections is dense and is substantially perpendicular to the detector.
24. The system of claim 22 , wherein the detection process further causes the means for processing, when establishing the motion model, to establish the motion model where the data item projections substantially converge in the parameter space with a cell in the voting array associated with a global maximum.