Digital signal processing methods, systems and computer program products that identify threshold positions and values

• US 20060098845A1
• Filed: 05/20/2005
• Published: 05/11/2006
• Est. Priority Date: 11/05/2004
• Status: Active Grant

• Current Assignee: Euripides Sotiriades, Kyprianos Papademetriou
• Original Assignee: Euripides Sotiriades, Kyprianos Papademetriou
• Inventors: Sotriropoulos, Stamatios, Dollas, Apostolos, Papademetriou, Kyprianos
• Granted Patent: US 7,583,819 B2
• Time in Patent Office: Days
• US Class Current: 382/107
• CPC Class Codes: G06F 18/00 Pattern recognition
• CPC Class Codes: G06F 2218/08 Feature extraction
• CPC Class Codes: G06V 40/20 Movements or behaviour, e.g. gesture recognition recognition of facial expressions G06V40/16

Abstract

A signal is sampled and the samples are divided into windows. One of at least three values is associated with a respective window based on whether the samples in the window are relatively constant, incremental or decremental, to thereby generate an approximation vector. Locations of thresholds in the approximation vector are determined, to generate a threshold position vector. Threshold values are assigned to the threshold positions in the threshold position vector. The threshold values, taken alone or in combination with a threshold position vector, may be used to characterize the signal. A training sequence may be used to adapt the functionality to individual needs.

Claims

1. A method of digitally processing a signal comprising:

sampling the signal and dividing the samples into windows;

associating one of at least three values with a respective window based upon whether the samples in the window are relatively constant, incremental, or decremental, to thereby generate an approximation vector;

determining locations of thresholds in the approximation vector to generate a threshold position vector that indicates the locations of thresholds in the approximation vector; and

assigning threshold values to the threshold positions in the threshold position vector to thereby generate a set of threshold values.

2. A method according to claim 1 wherein the signal comprises an acceleration signal that is generated by motion and wherein the threshold position vector and the set of threshold values characterize the motion.

3. A method according to claim 1 wherein associating one of at least three values with a respective window comprises ignoring local extrema in the samples.

4. A method according to claim 1 wherein associating one of at least three values with a respective window comprises:

repeatedly comparing a previous sample in a window to a next sample in the window;

incrementing a respective one of three counters when the next sample is, respectively, about the same as, greater than or less than, the previous sample; and

associating the one of the three values with the respective window based upon a highest value of the three counters.

5. A method according to claim 1 wherein assigning threshold values to the threshold positions in the threshold position vector comprises:

assigning threshold values to the threshold positions in the threshold position vector based on a highest or lowest valued sample in the respective windows that correspond to the locations of the thresholds in the approximation vector, to thereby generate a set of threshold values.

6. A method according to claim 1 wherein associating one of at least three values with a respective window is preceded by:

obtaining first and second signals representative of a common state variable;

identifying one of the first or second signals as a primary signal and the other of the first or second signals as a secondary signal; and

wherein associating one of at least three values with a respective window, determining locations of thresholds in the approximation vector and assigning threshold values to the threshold positions in the threshold position vector are performed on the primary signal, but are not performed on the secondary signal.

7. A method according to claim 6 wherein identifying one of the first or second signals as a primary signal and the other of the first or second signals as a secondary signal comprises:

identifying one of the first or second signals as the primary signal based on peak-to-peak excursions thereof.

8. A method according to claim 2 wherein associating one of at least three values with a respective window is preceded by:

obtaining first and second acceleration signals from respective first and second acceleration axes as a result of the motion;

identifying one of the first or second acceleration signals as a primary acceleration signal and the other of the first or second acceleration signals as a secondary acceleration signal; and

wherein associating one of at least three values with a respective window, determining locations of thresholds in the approximation vector and assigning threshold values to the threshold positions in the threshold position vector are performed on the primary acceleration signal, but are not performed on the secondary acceleration signal, to characterize the motion.

9. A method according to claim 8 wherein identifying one of the first or second acceleration signals as a primary acceleration signal and the other of the first or second acceleration signals as a secondary acceleration signal comprises:

identifying one of the first or second acceleration signals as the primary acceleration signal based on peak-to-peak excursions thereof.

10. A method according to claim 1:

wherein sampling the signal and dividing the samples into windows, associating one of at least three values with a respective window, determining locations of thresholds in the approximation vector and assigning threshold values to the threshold positions in the threshold position vector are performed repeatedly for a plurality of signals to obtain a threshold position vector and at least one corresponding set of threshold values that characterize the plurality of signals;

wherein sampling the signal and dividing the samples into windows, associating one of at least three values with a respective window, determining locations of thresholds in the approximation vector and assigning threshold values to the threshold positions in the threshold position vector are performed for an unknown signal to obtain a threshold position vector and a set of threshold values that characterize the unknown signal; and

wherein the method further comprises comparing the at least one threshold position vector and the at least one corresponding set of threshold values that characterize the plurality of signals and the threshold position vector and the set of threshold values that characterize the unknown signal to attempt to identify the unknown signal as one of the plurality of signals.

11. A method according to claim 2:

wherein sampling the signal and dividing the samples into windows, associating one of at least three values with a respective window, determining locations of thresholds in the approximation vector and assigning threshold values to the threshold positions in the threshold position vector are performed repeatedly for a plurality of acceleration signals that result from repeatedly performing a given motion to obtain a threshold position vector and at least one corresponding set of threshold values that characterize the given motion;

wherein sampling the signal and dividing the samples into windows, associating one of at least three values with a respective window, determining locations of thresholds in the approximation vector and assigning threshold values to the threshold positions in the threshold position vector are performed for an unknown acceleration signal that results from performing an unknown motion to obtain a threshold position vector and a set of threshold values that characterize the unknown motion; and

wherein the method further comprises comparing the at least one threshold position vector and the at least one corresponding set of threshold values that characterize the given motion and the threshold position vector and the set of threshold values that characterize the unknown motion to attempt to identify the unknown motion as one of the given motions.

12. A method according to claim 1:

wherein sampling the signal and dividing the samples into windows, associating one of at least three values with a respective window, determining locations of thresholds in the approximation vector and assigning threshold values to the threshold positions in the threshold position vector are performed repeatedly for a plurality of signals to obtain a threshold position vector and at least one corresponding set of threshold values that characterize the plurality of signals.

13. A method according to claim 2:

wherein sampling the signal and dividing the samples into windows, associating one of at least three values with a respective window, determining locations of thresholds in the approximation vector and assigning threshold values to the threshold positions in the threshold position vector are performed repeatedly for a plurality of acceleration signals that result from repeatedly performing a given motion to obtain a threshold position vector and at least one corresponding set of threshold values that characterize the given motion.

14. A method according to claim 1 wherein determining locations of thresholds in the approximation vector to generate a threshold position vector that indicates the locations of thresholds in the approximation vector comprises:

repeatedly comparing preceding and succeeding windows of the approximation vector to identify sets of increasing values that are followed by sets of decreasing values and/or sets of decreasing values that are followed by sets of increasing values.

15. A method according to claim 14 wherein repeatedly comparing preceding and succeeding windows of the approximation vector is repeatedly performed until a predetermined number of sets of increasing values that are followed by sets of decreasing values and/or sets of decreasing values that are followed by sets of increasing values is identified.

16. A method according to claim 14 wherein repeatedly comparing preceding and succeeding windows of the approximation vector is repeatedly performed until one less than a predetermined number of sets of increasing values that are followed by sets of decreasing values and/or sets of decreasing values that are followed by sets of increasing values is identified, the method further comprising identifying a single intermediate threshold.

17. A method according to claim 2 wherein determining locations of thresholds in the approximation vector to generate a threshold position vector that indicates the locations of thresholds in the approximation vector comprises:

repeatedly comparing preceding and succeeding windows of the approximation vector to identify sets of accelerations that are followed by sets of decelerations and/or sets of decelerations that are followed by sets of accelerations.

18. A method according to claim 17 wherein repeatedly comparing preceding and succeeding windows of the approximation vector is repeatedly performed until a predetermined number of sets of accelerations that are followed by sets of decelerations and/or sets of decelerations that are followed by sets of accelerations is identified.

19. A method according to claim 17 wherein repeatedly comparing preceding and succeeding windows of the approximation vector is repeatedly performed until one less than a predetermined number of sets of accelerations that are followed by sets of decelerations and/or sets of decelerations that are followed by sets of accelerations is identified, the method further comprising identifying a single intermediate threshold.

20. A digital signal processing system that is configured to perform a method according to claim 1.

21. A computer program product that is configured to perform a method according to claim 1.

22. A motion detection system comprising:

a motion sensor that is configured to generate an acceleration signal in response to motion thereof; and

digital signal processing means comprising;

means for sampling the acceleration signal and dividing the samples into windows;

means for associating one of at least three values with a respective window based upon whether the samples in the window are relatively constant, incremental, or decremental, to thereby generate an approximation vector;

means for determining locations of thresholds in the approximation vector to generate a threshold position vector that indicates the locations of thresholds in the approximation vector; and

means for assigning threshold values to the threshold positions in the threshold position vector to thereby generate a set of threshold values.

23. A data structure that characterizes a signal comprising:

an approximation vector that associates one of at least three values with a respective window of samples of the signal based upon whether the samples in the window are relatively constant, incremental, or decremental;

a threshold position vector that indicates locations of thresholds in the approximation vector; and

a threshold value vector that identifies threshold values for the threshold positions in the threshold position vector.