High-frame-rate real-time multiscale spatiotemporal disparity on distributed low-power event-based neuromorphic hardware
High-framerate real-time spatiotemporal disparity mechanisms on neuromorphic hardware are provided. In various embodiments, a first and second spiking input sensor each output a time series of spikes corresponding to a plurality of frames. A neurosynaptic network is configured to receive the time series of spikes corresponding to the plurality of frames; accumulate the time series of spikes in a ring buffer, thereby creating a plurality of temporal scales; for each corresponding pair of frames from the first and second spiking input sensors, determining a mapping of pixels in one of the pair of frames to pixels in the other of the pair of frames based on similarity; based on the pixel mapping, determining a disparity map.
1. A system comprising:
a first and second spiking input sensor, each outputting a time series of spikes corresponding to a plurality of frames;
a neurosynaptic network configured to:
receive the time series of spikes corresponding to the plurality of frames;
accumulate the time series of spikes in a ring buffer, thereby creating a plurality of temporal scales;
for each corresponding pair of frames from the first and second spiking input sensors, determining a mapping of pixels in one of the pair of frames to pixels in the other of the pair of frames based on similarity;
based on the pixel mapping, determining a disparity map.
2. The system of claim 1 , wherein the input sensors comprise image sensors.
3. The system of claim 1 , wherein the neurosynaptic network is further configured to denoise the plurality of frames.
4. The system of claim 2 , wherein said denoising comprises morphological erosion/dilation.
5. The system of claim 1 , wherein the neurosynaptic network further comprises a multiplexor adapted to divide each of the plurality of frames into a plurality of non-overlapping patches.
6. The system of claim 1 , wherein the neurosynaptic network is further configured to denoise the disparity map.
7. The system of claim 6 , wherein said denoising comprises morphological erosion/dilation.
8. The system of claim 1 , wherein the neurosynaptic network comprises a plurality of chips.
9. The system of claim 8 , wherein each chip is adapted to determine the disparity map for a non-overlapping patch of each of the plurality of frames.
10. The system of claim 8 , wherein the neurosynaptic network is further adapted to fuse the disparity map from each chip into a global disparity map.
11. The system of claim 8 , further comprising a computing node adapted to fuse the disparity map from each chip into a global disparity map.
12. The system of claim 1 , wherein the neurosynaptic network is further adapted to rectify the plurality of frames.
13. The system of claim 1 , the neurosynaptic network is further adapted to verify the pixel mapping by determining a reverse pixel mapping.
14. A method comprising:
receive at a neurosynaptic network time series of spikes corresponding to a plurality of frames from a first and second spiking input sensor;
accumulating the time series of spikes in a ring buffer of the neurosynaptic network, thereby creating a plurality of temporal scales;
for each corresponding pair of frames from the first and second spiking input sensors, determining by the neuro synaptic network a mapping of pixels in one of the pair of frames to pixels in the other of the pair of frames based on similarity;
based on the pixel mapping, determining by the neurosynaptic network a disparity map.
15. The method of claim 14 , wherein the input sensors comprise image sensors.
16. The method of claim 14 , wherein the neurosynaptic network is further configured to denoise the plurality of frames.
17. The method of claim 15 , wherein said denoising comprises morphological erosion/dilation.
18. The method of claim 14 , wherein the neurosynaptic network further comprises a multiplexor adapted to divide each of the plurality of frames into a plurality of non-overlapping patches.
19. The method of claim 14 , wherein the neurosynaptic network is further configured to denoise the disparity map.
20. The method of claim 19 , wherein said denoising comprises morphological erosion/dilation.