Methods and systems for data processing and their applications
A method ( 400 ) of processing signal outputs of a plurality of topologically distinct sensors in response to stimuli is described. The method comprises obtaining ( 402 ) a plurality of temporal sensor outputs in parallel. Thereafter, features are extracted ( 406 ), the features having dynamic behavior pattern. The extraction is performed in a topology consistent way by arithmetic processing in parallel of neighboring temporal sensor outputs. Furthermore, a quality of the extracted features is being determined.
1. A method of processing signal outputs of a plurality of topologically distinct sensors in response to mechanical stimuli, the method comprising:
obtaining a plurality of topologically arranged temporal sensor outputs in parallel,
extracting features by arithmetic processing in parallel of neighbouring temporal sensor outputs in a topology consistent way based on spatial filtering, the features being identified by a dynamic behaviour pattern, and
determining a quality of the extracted features.
2. A method according to claim 1 , wherein extracting features taking in a topology consistent way comprises arranging the features in a topology maintaining array or arrays.
3. A method according to claim 1 , wherein obtaining a plurality of temporal sensor outputs in parallel comprises sampling a plurality of sensor outputs in time to generate the temporal sensor outputs.
4. A method according to claim 1 , wherein the temporal sensor outputs are sensor outputs from mechanical sensors or tactile sensors.
5. A method according to claim 1 , wherein arithmetic processing comprises processing said neighbouring temporal sensor outputs in the temporal domain.
6. A method according to claim 1 , wherein said extracting of features in a topology consistent way comprises recognising the features in sensor outputs and preserving their spatial localisation.
7. A method according to claim 2 , the method further comprising spatially filtering or spatio-temporal filtering of the array or arrays.
8. A method according to claim 7 , wherein said spatially filtering comprises extraction of an event based on a plurality of neighbouring extracted features in said topology maintaining array or arrays.
9. A method according to claim 1 , either or both wherein the arithmetic processing comprises at least one of:
applying a bandpass, low-pass or high-pass filter or a combination of bandpass, low-pass or high-pass filters for temporal feature extraction,
using a cellular non-linear network,
only adding, subtracting, delaying, thresholding, logical or morphological functions, and
applying a filter having a single filter input signal and resulting in multiple filter output signal, wherein each filter output signal is based on the subtraction of a modified input signal from an original input signal; or
wherein the arithmetic processing in parallel of neighbouring temporal sensor outputs comprises arithmetic processing in parallel of only part of the neighbouring temporal sensor outputs.
10. A method according to claim 1 , wherein the dynamic behaviour comprises any of a variation, vibration or oscillation.
11. A method according to claim 1 , wherein said quality of the extracted features comprises a stability of the dynamic behaviour among a plurality of neighbours.
12. A method according to claim 1 , wherein the method processes said sensor outputs to improve active or passive perception of events or features occurring in the environment.
13. A method for controlling a computerised system comprising at least one movable element comprising a plurality of topologically distinct sensors, the method comprising;
measuring a plurality of output signals from said mechanical sensors,
processing said plurality of output signals using the steps of the method according to of claim 1 , and
controlling said at least one movable element based on said determined quality of the extracted features.
14. A data processor for processing signal outputs of a plurality of topologically distinct sensors in response to mechanical stimuli, the processor comprising:
an input configured to receive a plurality of topologically arranged temporal sensor outputs in parallel,
a feature extractor configured to extract features with a dynamical behaviour pattern in a topology consistent way based on spatial filtering, the feature extractor comprising an arithmetic processor configured to perform arithmetic processing in parallel neighbouring temporal samples of the sensor outputs, and
a determinator configured to determine a quality of the extracted features.
15. A data processor according to claim 14 , wherein the input configured to receive a plurality of temporal sensor outputs in parallel comprises a sampler configured to sample a plurality of sensor outputs in time to generate the temporal sensor outputs.
16. A data processor according to claim 14 , wherein the arithmetic processor is configured to perform arithmetic processing of said neighbouring temporal sensor outputs in the temporal domain.
17. A data processor according to claim 14 , the data processor is configured to arrange the extracted features in a topology maintaining array or arrays, or
the data processor is configured to recognize features in sensor outputs and preserve their spatial localisation, or
the data processor is configured to arrange the extracted features in a topology maintaining array or arrays and comprising a filter for spatially or spatio-temporal filtering of the array or arrays.
18. A data processor according to claim 17 , wherein the filter is configured to extract an event based on a plurality of neighbouring extracted features in said topology maintaining array or arrays.
19. A data processor according to claim 14 , either or both wherein the arithmetic processor comprises at least one of:
a bandpass filter or a combination of bandpass filters for extracting temporal features,
a cellular neural network or a cellular non-linear network,
a filter based on only substracting, delaying, thresholding, logical or morphological functions, or
a filter having a single filter input signal and resulting in multiple filter output signal, wherein each filter output signal is based on the subtraction of a modified input signal from an original input signal; or
wherein the arithmetic processor for processing in parallel neighbouring temporal sensor outputs is an arithmetic processor that arithmetically processes in parallel only part of the neighbouring temporal sensor outputs.
20. A data processor according to claim 14 , wherein the dynamic behaviour comprises any of a variation, vibration, or oscillation.
21. A data processor according to claim 14 , wherein said quality of the extracted features comprises a stability of the dynamic behaviour among a plurality of neighbours.
22. A computerised system comprising at least one movable element comprising a plurality of topologically distinct sensors, and a data processor according to claim 14 .
23. A computerised system according to claim 22 , the system further comprising a controller for controlling the at least one moveable element as function of the obtained quality of the extracted features.
24. A non-transitory computer program product comprising instructions for processing signal outputs of a plurality of topologically distinct mechanical sensors in response to mechanical stimuli, the computer program product, when executed on a computer, executing the method according to claim 1 .
25. A non-transitory machine-readable data storage device storing the computer program product recited in claim 24 .