Sonar and Radar SAR Processing for Parking Lot Detection

Transcript

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing Experiments Conclusions Sonar and Radar SAR processing for Parking Lot Detection James Mure-Dubois, Fran¸ cois Vincent, David Bonacci IRS 2011 - 08.09.2011 J. Mure-Dubois/ 08.09.2011 - 1 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction Problem deﬁnition Motivation for SAR processing Radar and Sonar systems SAR processing Experiments Conclusions Problem deﬁnition - Parking Lot Detection Main idea Realize new functions with existing 77GHz radar sensors (cf. [WZ09, WZ10]) Parking Lot Detection function As the carrier vehicle moves, a radar maps the area surrounding the vehicle. J. Mure-Dubois/ 08.09.2011 - 2 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction Problem deﬁnition Motivation for SAR processing Radar and Sonar systems SAR processing Experiments Conclusions Problem deﬁnition - Parking Lot Detection Main idea Realize new functions with existing 77GHz radar sensors (cf. [WZ09, WZ10]) Parking Lot Detection function As the carrier vehicle moves, a radar maps the area surrounding the vehicle. The collected radar data is processed to compose a high resolution map. The map allows identifying suitable parking spots. J. Mure-Dubois/ 08.09.2011 - 2 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction Problem deﬁnition Motivation for SAR processing Radar and Sonar systems SAR processing Experiments Conclusions Motivation for SAR processing v x0 y0 y x High cross-range resolution requires large antennas Embedded radars must have small dimensions Synthetic Aperture Radar (SAR) allows emulating a large antenna with a smaller sensor SAR processing requires coherent integration for the whole time where the target passes through the antenna beam J. Mure-Dubois/ 08.09.2011 - 3 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction Problem deﬁnition Motivation for SAR processing Radar and Sonar systems SAR processing Experiments Conclusions Motivation for SAR processing x0 y0 y x High cross-range resolution requires large antennas Embedded radars must have small dimensions Synthetic Aperture Radar (SAR) allows emulating a large antenna with a smaller sensor SAR processing requires coherent integration for the whole time where the target passes through the antenna beam J. Mure-Dubois/ 08.09.2011 - 3 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction Problem deﬁnition Motivation for SAR processing Radar and Sonar systems SAR processing Experiments Conclusions Motivation for SAR processing v x0 y0 y x High cross-range resolution requires large antennas Embedded radars must have small dimensions Synthetic Aperture Radar (SAR) allows emulating a large antenna with a smaller sensor SAR processing requires coherent integration for the whole time where the target passes through the antenna beam J. Mure-Dubois/ 08.09.2011 - 3 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction Problem deﬁnition Motivation for SAR processing Radar and Sonar systems SAR processing Experiments Conclusions Motivation for SAR processing Synthetic Array Length x0 y0 y x High cross-range resolution requires large antennas Embedded radars must have small dimensions Synthetic Aperture Radar (SAR) allows emulating a large antenna with a smaller sensor SAR processing requires coherent integration for the whole time where the target passes through the antenna beam J. Mure-Dubois/ 08.09.2011 - 3 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction Problem deﬁnition Motivation for SAR processing Radar and Sonar systems SAR processing Experiments Conclusions Radar and Sonar systems Radar system Prototype based on a commercial autmotive CW radar Operation around 77GHz → λ = 3.9mm Ultrasound sonar system Assembled for teaching activities at ISAE [VMC+07] Uses a conﬁgurable pulse waveform Total motion during SAS measurement : 3m λ = 8.5mm J. Mure-Dubois/ 08.09.2011 - 4 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing SAR hyperbolic geometry Spatio-temporal stripmap SAR Range-Doppler algorithm with RCC Monochromatic Range-Doppler algorithm Experiments Conclusions SAR hyperbolic geometry r(x)2 = (x − x0)2 + y2 0 In the radar’s frame of reference, each point target follows an hyperbolic path : the PSF is 2D h (x − x0, r; y0) = e −j 4π c (fc +α r c ) √ y2 0 +(x−x0)2 Hyperbolas curvature gets higher as targets move closer to the radar’s path Range curvature compensation is important for scenes with wide range swath. J. Mure-Dubois/ 08.09.2011 - 5 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing SAR hyperbolic geometry Spatio-temporal stripmap SAR Range-Doppler algorithm with RCC Monochromatic Range-Doppler algorithm Experiments Conclusions SAR hyperbolic geometry r(x)2 = (x − x0)2 + y2 0 In the radar’s frame of reference, each point target follows an hyperbolic path : the PSF is 2D h (x − x0, r; y0) = e −j 4π c (fc +α r c ) √ y2 0 +(x−x0)2 Hyperbolas curvature gets higher as targets move closer to the radar’s path Range curvature compensation is important for scenes with wide range swath. J. Mure-Dubois/ 08.09.2011 - 5 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing SAR hyperbolic geometry Spatio-temporal stripmap SAR Range-Doppler algorithm with RCC Monochromatic Range-Doppler algorithm Experiments Conclusions Spatio-temporal stripmap SAR Range Compression sr (x) sy1 (x) Bin select Bin write PSF estim. h* (x,r(x,y1)) h* (x,r(x,y0)) sr (x) s (x,r) sy0 (x) I (x,y) PSF estim. Bin select Bin write Azimuth Compression Azimuth compression requires to estimate the PSF for each position and range bin The matched ﬁlter is applied to each range bin individually J. Mure-Dubois/ 08.09.2011 - 6 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing SAR hyperbolic geometry Spatio-temporal stripmap SAR Range-Doppler algorithm with RCC Monochromatic Range-Doppler algorithm Experiments Conclusions Range-Doppler algorithm with RCC Range Compression Azimuth FFT Azimuth IFFT H* (Kx,r(x0,y)) S (Kx,r) s (x,r) Sm(Kx,y) I (x,y) PSF estim. Azimuth Compression The PSF is estimated for a reference position but for each range bin. The matched ﬁlter is applied to each range bin individually. J. Mure-Dubois/ 08.09.2011 - 7 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing SAR hyperbolic geometry Spatio-temporal stripmap SAR Range-Doppler algorithm with RCC Monochromatic Range-Doppler algorithm Experiments Conclusions Monochromatic Range-Doppler algorithm Range Compression Azimuth FFT Azimuth IFFT H* (Kx,r(x0,y0)) Azimuth Compression S (Kx,r) s (x,r) Sm(Kx,y) I (x,y) PSF estim. Azimuth Compression The PSF is estimated for a single reference position and range. The same matched ﬁlter is applied to all range bins (memory savings in embedded application). J. Mure-Dubois/ 08.09.2011 - 8 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing Experiments SAS system SAS PSF SAS for extended objects 77GHz SAR system Conclusions Ultrasound SAS system Tests Small target (PSF) Empty parking place Occupied parking place J. Mure-Dubois/ 08.09.2011 - 9 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing Experiments SAS system SAS PSF SAS for extended objects 77GHz SAR system Conclusions Ultrasound SAS system PSF Range Compression Azimuth FFT Azimuth IFFT H* (Kx,r(x0,y)) S (Kx,r) s (x,r) Sm(Kx,y) I (x,y) PSF estim. Azimuth Compression Range Compression sr (x) sy1 (x) Bin select Bin write PSF estim. h* (x,r(x,y1)) h* (x,r(x,y0)) sr (x) s (x,r) sy0 (x) I (x,y) PSF estim. Bin select Bin write Azimuth Compression J. Mure-Dubois/ 08.09.2011 - 10 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing Experiments SAS system SAS PSF SAS for extended objects 77GHz SAR system Conclusions Ultrasound SAS system PSF Range Compression Azimuth FFT Azimuth IFFT H* (Kx,r(x0,y0)) Azimuth Compression S (Kx,r) s (x,r) Sm(Kx,y) I (x,y) PSF estim. Azimuth Compression Range Compression sr (x) sy1 (x) Bin select Bin write PSF estim. h* (x,r(x,y1)) h* (x,r(x,y0)) sr (x) s (x,r) sy0 (x) I (x,y) PSF estim. Bin select Bin write Azimuth Compression J. Mure-Dubois/ 08.09.2011 - 10 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing Experiments SAS system SAS PSF SAS for extended objects 77GHz SAR system Conclusions Monochromatic Range-Doppler defocus Range Compression Azimuth FFT Azimuth IFFT H* (Kx,r(x0,y0)) Azimuth Compression S (Kx,r) s (x,r) Sm(Kx,y) I (x,y) PSF estim. Azimuth Compression Range Compression sr (x) sy1 (x) Bin select Bin write PSF estim. h* (x,r(x,y1)) h* (x,r(x,y0)) sr (x) s (x,r) sy0 (x) I (x,y) PSF estim. Bin select Bin write Azimuth Compression J. Mure-Dubois/ 08.09.2011 - 11 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing Experiments SAS system SAS PSF SAS for extended objects 77GHz SAR system Conclusions SAS for extended objects Objects boundaries are not clearly deﬁned in raw data J. Mure-Dubois/ 08.09.2011 - 12 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing Experiments SAS system SAS PSF SAS for extended objects 77GHz SAR system Conclusions SAS for extended objects Range Compression Azimuth FFT Azimuth IFFT H* (Kx,r(x0,y)) S (Kx,r) s (x,r) Sm(Kx,y) I (x,y) PSF estim. Azimuth Compression Range Compression Azimuth FFT Azimuth IFFT H* (Kx,r(x0,y)) S (Kx,r) s (x,r) Sm(Kx,y) I (x,y) PSF estim. Azimuth Compression Objects boundaries are not clearly deﬁned in raw data SAS image is sharper, allows to separate objects J. Mure-Dubois/ 08.09.2011 - 12 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing Experiments SAS system SAS PSF SAS for extended objects 77GHz SAR system Conclusions Monochromatic SAS focusing, extended scene When using monochromatic methods, focus distance has a strong inﬂuence on map sharpness. J. Mure-Dubois/ 08.09.2011 - 13 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing Experiments SAS system SAS PSF SAS for extended objects 77GHz SAR system Conclusions Monochromatic SAS focusing, extended scene Range Compression Azimuth FFT Azimuth IFFT H* (Kx,r(x0,y0)) Azimuth Compression S (Kx,r) s (x,r) Sm(Kx,y) I (x,y) PSF estim. Azimuth Compression When using monochromatic methods, focus distance has a strong inﬂuence on map sharpness. J. Mure-Dubois/ 08.09.2011 - 13 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing Experiments SAS system SAS PSF SAS for extended objects 77GHz SAR system Conclusions Monochromatic SAS focusing, extended scene Range Compression Azimuth FFT Azimuth IFFT H* (Kx,r(x0,y)) S (Kx,r) s (x,r) Sm(Kx,y) I (x,y) PSF estim. Azimuth Compression When using monochromatic methods, focus distance has a strong inﬂuence on map sharpness. When range curvature is compensated for, the output image is as sharp as an image produced with the slow space-time SAR method J. Mure-Dubois/ 08.09.2011 - 13 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing Experiments SAS system SAS PSF SAS for extended objects 77GHz SAR system Conclusions Monochromatic SAS focusing, extended scene Range Compression sr (x) sy1 (x) Bin select Bin write PSF estim. h* (x,r(x,y1)) h* (x,r(x,y0)) sr (x) s (x,r) sy0 (x) I (x,y) PSF estim. Bin select Bin write Azimuth Compression When using monochromatic methods, focus distance has a strong inﬂuence on map sharpness. When range curvature is compensated for, the output image is as sharp as an image produced with the slow space-time SAR method J. Mure-Dubois/ 08.09.2011 - 13 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing Experiments SAS system SAS PSF SAS for extended objects 77GHz SAR system Conclusions 77GHz radar system Objects boundaries are diﬃcult to locate in raw data J. Mure-Dubois/ 08.09.2011 - 14 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing Experiments SAS system SAS PSF SAS for extended objects 77GHz SAR system Conclusions 77GHz radar system Range Compression Azimuth FFT Azimuth IFFT H* (Kx,r(x0,y)) S (Kx,r) s (x,r) Sm(Kx,y) I (x,y) PSF estim. Azimuth Compression Objects boundaries are diﬃcult to locate in raw data SAR image brings a signiﬁcant sharpness improvement J. Mure-Dubois/ 08.09.2011 - 14 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing Experiments Conclusions Conclusions Bibliography Conclusions Range Compression Azimuth FFT Azimuth IFFT H* (Kx,r(x0,y)) S (Kx,r) s (x,r) Sm(Kx,y) I (x,y) PSF estim. Azimuth Compression Summary Two low complexity SAR processing methods for parking lot detection were compared : Range-Doppler with range curvature compensation Monochromatic range-Doppler Ultrasound SAS system allowed to quantify PSF degradation in monochromatic range-Doppler. Results were validated on a 77GHz CW radar prototype, on extended targets. Possible improvements Study robustness against errors introduced by radar carrier motion J. Mure-Dubois/ 08.09.2011 - 15 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing Experiments Conclusions Conclusions Bibliography Bibliography I F. Vincent, B. Mouton, E. Chaumette, C. Nouals, and O. Besson. Synthetic aperture radar demonstration kit for signal processing education. In Acoustics, Speech and Signal Processing, 2007. ICASSP 2007. IEEE International Conference on, volume 3, pages III–709 –III–712, april 2007. Huaming Wu and T. Zwick. Automotive sar for parking lot detection. In German Microwave Conference, 2009, pages 1 –8, mar 2009. J. Mure-Dubois/ 08.09.2011 - 16 -

Sonar and Radar SAR processing for Parking Lot Detection J.

Mure-Dubois Introduction SAR processing Experiments Conclusions Conclusions Bibliography Bibliography II Huaming Wu and T. Zwick. A novel motion compensation algorithm for automotive sar : Simulations and experiments. In German Microwave Conference, 2010, pages 222 –226, Berlin, March 2010. J. Mure-Dubois/ 08.09.2011 - 17 -

Sonar and Radar SAR Processing for Parking Lot ...

Sonar and Radar SAR Processing for Parking Lot Detection

Other Decks in Research

Featured

Transcript