초록 ▼

An attitude estimator provides non-causal attitude estimates for real-time motion compensation of sensed images on a moving platform. A non-causal filter processes uncompensated attitude samples received with a latency from an IMU at a high rate clock to provide an attitude estimate that is strictly

An attitude estimator provides non-causal attitude estimates for real-time motion compensation of sensed images on a moving platform. A non-causal filter processes uncompensated attitude samples received with a latency from an IMU at a high rate clock to provide an attitude estimate that is strictly non-causal but satisfies a just-in-time (JIT) criteria for real-time motion compensation of images captured at a low rate clock. On-average the error of the non-causal attitude estimate is less than the error of a causal attitude estimate. If the lag added by the non-causal filter is greater than the latency, the effective transfer function of the non-causal filter has a negative gain slope that attenuates high frequency noise of the uncompensated attitude samples. The attitude estimator may also include a causal filter to generate a causal attitude estimate for real-time active stabilization of the image sensor at the high rate clock.

대표청구항 ▼

1. An attitude estimator for providing attitude estimates for real-time motion compensation of images on a moving platform, wherein an IMU provides uncompensated attitude samples S(hT) of the moving platform at a high rate clock with a sampling period T and a dock index h, said samples S(hT) receive

1. An attitude estimator for providing attitude estimates for real-time motion compensation of images on a moving platform, wherein an IMU provides uncompensated attitude samples S(hT) of the moving platform at a high rate clock with a sampling period T and a dock index h, said samples S(hT) received with a latency L, an image sensor captures images O(kMT) at a low rate clock with a sampling period MT where M is greater than one and a clock index k, each image captured over an integration period having a center of integration (COI) at the center of the integration period synchronized to the low rare clock followed by a readout period, and a motion compensator that initiates real-time motion compensation on the captured image I(kMT) at the end of the readout period at a just-in-time constraint J measured in a number of high rate clock samples equal to one half the integration period plus the readout period after the COI, the attitude estimator comprising: a non-causal sampling clock corresponding to the low rate clock kMT delayed by N high rate clock samples where N is less than or equal to J; anda non-causal filter that filters the uncompensated attitude samples S(hT) including at least one sample received after the COI to provide a non-causal attitude estimate A(kMT|(kM+N)T) for the previous COI that is delayed by N−L high rate samples and synchronized to the non-causal sampling clock, said motion compensator using the non-causal attitude estimate A(kMT|(kM+N)T) to initiate motion compensation on the captured image I(kMT). 2. The attitude estimator of claim 1, wherein the non-causal attitude estimate A(kMT|(kM+N)T) is synchronized to the high rate clock with an error of less than one high rate clock cycle. 3. The attitude estimator of claim 1, wherein the uncompensated attitude samples S(hT) include errors due to measurement noise and the latency L, wherein on-average the non-causal attitude estimate A(kMT|(kM+N)T) includes less error than a causal attitude estimate A(kMT|kMT) which includes less error than the measured attitude estimate S(kMT) at current time sample kMT. 4. The attitude estimator of claim 1, wherein the JIT constraint J is greater than the latency L, said non-causal filter interpolating multiple uncompensated attitude samples to provide the non-causal attitude estimate A(kMT|(kM+N)T). 5. The attitude estimator of claim 4, where the delay N is greater than the latency L, said non-causal filter having an effective transfer function having a negative gain slope that attenuates high frequency noise of the uncompensated attitude samples. 6. The attitude estimator of claim 5, wherein the integration period is fixed and the delay N is equal to the JIT constraint J. 7. The attitude estimator of claim 6, wherein the integration period is variable and the delay N is no greater than the readout period. 8. The attitude estimator of claim 1, wherein the delay N equals the latency L, said non-causal filter sampling the uncompensated attitude sample at S(hT−NT) to provide the non-causal attitude estimate A(kMT|(kM+N)T). 9. The attitude estimator of claim 1, where the JIT constraint J is less than the latency L, said non-causal filter extrapolating from multiple uncompensated attitude samples to provide the a non-causal attitude estimate A(kMT|(kM+N)T). 10. The attitude estimator of claim 1, wherein the non-causal filter comprises: a low-pass filter that adds N−L high rate clock samples of lag to provide an estimate at the high rate clock; anda sampler that samples the estimate at the non-causal sampling clock to provide the non-causal attitude estimate A(kMT|(kM+N)T). 11. The attitude estimator of claim 1, wherein the non-causal filter comprises: a buffer of width W greater than N that receives the uncompensated attitude samples S(hT) with latency L;an attitude trend filter that shifts the uncompensated attitude samples in the buffer by latency L and fits a continuous function f(t) to the shifted uncompensated attitude samples at the high rate clock; anda sampler that samples the continuous function f(t) at the non-causal sampling clock to provide the non-causal attitude estimate A(kMT|(kM+N)T). 12. The attitude estimator of claim 1, wherein a stabilizer actively stabilizes attitude errors in orientation of the image sensor in real-time at the high rate clock, further comprising: a causal filter that filters only uncompensated attitude samples received at or prior to hT of the high rate clock to provide a causal attitude estimate A(hT|hT) synchronized to the high rate clock, said stabilizer using the causal attitude estimate A(hT|hT) to actively stabilize the orientation of the image sensor at the high rate clock. 13. The attitude estimator of claim 12, wherein the uncompensated attitude samples include errors due to measurement noise and the latency L of receiving the samples, wherein on-average the non-causal attitude estimate A(kMT|(kM+N)T) includes less error than the causal attitude estimate A(hT|hT) which includes less error than the measured attitude estimate S(hT) at current time sample hT. 14. The attitude estimator of claim 12, wherein the non-causal filter comprises a low-pass filter that adds N−L high rate clock samples of lag to provide a first estimate at the high rate clock and a first sampler that samples the first estimate at the non-causal sampling clock to provide the non-causal attitude estimate A(kMT|(kM+N)T) and the causal filter comprises a lead filter that adds L high rate clock samples of lead to provide a second estimate at the high rate clock and a second sampler that samples the second estimate at the high rate clock to provide the causal attitude estimate A(hT|hT). 15. The attitude estimator of claim 12, wherein the non-causal and causal filters comprises: a buffer of width W greater than N that receives the uncompensated attitude samples with latency L;an attitude trend filter that shifts the uncompensated attitude samples in the buffer by latency L and fits a continuous function f(t) to the shifted uncompensated attitude samples at the high rate clock;a first sampler that samples the continuous function f(t) at the non-causal sampling clock to provide the non-causal attitude estimate A(kMT|(kM+N)T); anda second sampler that samples the continuous function f(t) at the high rate clock to provide the causal attitude estimate A(hT|hT). 16. An attitude estimator for providing attitude estimates for real-time motion compensation of images on a moving, platform and for real-time active stabilization of the platform, wherein an IMU provides uncompensated attitude samples S(hT) of the moving platform at a high rate clock, with a sampling period T and a clock index h, said samples S(hT) received with a latency L, an image sensor captures images I(kMT) at a low rate clock with a sampling, period MT where M is greater than one and a clock, index k, each image captured over an integration period having a center of integration (COI) at the center of the integration period synchronized to the low rate clock followed by a readout period, and a motion compensator that initiates real-time motion compensation on the captured image I(kMT) at the end of the readout period at a just-in-time constraint J measured in a number of high rate clock samples equal to one half the integration period plus the readout period after the COI, and a stabilizer that actively stabilizes attitude errors in orientation of the image sensor in real-time at the high rate clock, the attitude estimator comprising: a causal filter that filters only uncompensated attitude samples received at or prior to hT of the high rate clock to provide a causal attitude estimate A(hT|hT) synchronized to the high rate clock, said stabilizer using the causal attitude estimate A(hT|hT) to actively stabilize the orientation of the image sensor at the high rate clock;a non-causal sampling clock corresponding to the low rate clock kMT delayed by N high rate clock samples where N is less than or equal to J; anda non-causal filter that filters the uncompensated attitude samples S(hT) including at least one sample received after the COI to provide a non-causal attitude estimate A(kMT|(kM+N)T) for the previous COI that is delayed by N−L high rate samples and synchronized to the non-causal sampling clock, said motion compensator using the non-causal attitude estimate A(kMT|(kM+N)T) to initiate motion compensation on the captured image I(kMT),wherein on-average the non-causal attitude estimate A(kMT|(kM+N)T) includes less error than the causal attitude estimate A(hT|hT), which includes less error than the measured attitude estimate S(hT) at current time sample hT. 17. The attitude estimator of claim 16, wherein the non-causal and causal filters comprises: a buffer of width W greater than N that receives the uncompensated attitude samples with latency L;an attitude trend filter that shifts the uncompensated attitude samples in the buffer by latency L and fits a continuous function f(t) to the shifted uncompensated attitude samples at the high rate clock;a first sampler that samples the continuous function f(t) at the non-causal sampling clock to provide the non-causal attitude estimate A(kMT|(kM+N)T); anda second sampler that samples the continuous function f(t) at the high rate clock to provide the causal attitude estimate A(hT|hT). 18. A system comprising: a moving platform;an IMU rigidly mounted on the moving platform, said IMU providing uncompensated attitude samples S(hT) of the moving platform at a high rate clock with a sampling period T and a clock index h;an image sensor that captures images I(kMT) at a low rate clock with a sampling period MT where M is greater than one and a clock, index k, each image captured over an integration period having a center of integration (COI) synchronized to the low rate clock followed by a readout period;a motion compensator that initiates real-time motion compensation on the captured image I(kMT) at the end of the readout period at a just-in-time (JIT) constraint J measured in a number of high rate clock samples equal to one half the integration period plus the readout period after the COI; andan attitude estimator that receives uncompensated attitude samples S(hT) with a latency L, said attitude estimator comprising: a non-causal sampling clock corresponding to the low rate clock kMT delayed by N high rate clock samples where N is less than or equal to J; anda non-causal filter that filters the uncompensated attitude samples S(hT) including at least one sample received after the COI to provide a non-causal attitude estimate A(kMT|(kM+N)T) for the previous COI that is delayed by N−L high rate samples and synchronized to the non-causal sampling clock at or before the JIT constraint J, said motion compensator using the non-causal attitude estimate A(kMT|(kM+N)T) to initiate motion compensation on the captured image I(kMT). 19. The system of claim 18, where the delay N is greater than the latency L, said non-causal filter having an effective transfer function having a negative gain slope that attenuates high frequency noise of the uncompensated attitude samples. 20. The system of claim 18, further comprising: a causal filter that filters only uncompensated attitude samples received at or prior to hT of the high rate clock to provide a causal attitude estimate A(hT|hT) synchronized to the high rate clock; anda stabilizer that uses the causal attitude estimate A(hT|hT) to actively stabilize the orientation of the image sensor at the high rate clock.