Okay, it's done :-). The RaySAR package can be downloaded on

https://github.com/StefanJAuer/RaySAR/releases

In there, the following components will be found:

- A readme to start with :-)

- A documentation of RaySAR (please read it carefully in advance of using RaySAR) and the methodological background (doctoral thesis)

- A license file related to the use of RaySAR (GNU v3)

- The RaySAR source code (additional POV-Ray code parts, MATLAB code) and a related guidance how to set up RaySAR

- Model files to test the functionality and properties of the RaySAR simulation environment

I'm looking forward to feedback from the community and new developments / research aspects based on RaySAR :-).

A dedicated Google Group for RaySAR improvements / applications can be found on

https://groups.google.com/forum/#!forum/raysar

Please take the opportunity to participate there in order to report your experiences and exchange ideas.

Good luck with your work.

Dear Stefan,

In principle, an accurate simulator like RaySAR could be very useful for the extraction of 3-D information in urban area by SAR stereogrammetric techniques. An initial DSM obtained co-registering the images could be determined in order to initialize the parameters of the simulator. One could then change the parameters of until a DSM is obtained which makes the simulated images radiometrically consistent with the SAR images pair (or pairs in case of multi-baseline approaches) at hand. A similar technique is discussed in the literature already (Brunner et al, 2010). Consistency here relates to the spatial distribution of intensity shadowing, layover, double-bounce, point-targets. Such an approach would present a prohibitive computational cost thus probably some gradient descent method in the simulator parameters space shall be preferred. This method could offer interesting results considering the sub-metric resolution of TerraSAR-X images acquired in the staring spotlight mode. 

Best regards,

Andrey

Dear Andrey,

RaySAR could be used in an iterative manner to find the best fitting geometry related to a scene using the core functions (ray tracing, SAR image simulation). Appropriate approximations would have to be made as long as the reconstructed geometry does not follow/represent the rule/grammar of the real object (usually: arrangement of physical existent corner reflectors or "ghost" corner reflectors). For instance, assuming a basic DSM with flat surfaces enables to indicate the trend which has been confirmed by several works (see Dominik's work as an example). Fitting the position of corner models may be a better alternative.

Thank you for your opinion :-).

Best regards

Stefan

Dear Stefan:

What an excited news it is! As you know, I read a lot about your work last year. The first glance I saw RaySAR, I found it a useful tool for 3D SAR imaging simulation, especially in complicated urban areas with known building models. As far as I am concerned, RaySAR contributes to my research in two aspects: (1) Restore the whole signal transmission process from SAR sensor to receiver, the interaction between the signal and the 3D building sence is my main focus which helps to further understand the nature of scatter, the origin of geometric distortions and multipath effect. (2) 3D analysis based on the amplitude and phase simulation results in SAR coordinate system. These information can be further utilized for scatter location, multi-bounce discrimination and so on.

If RaySAR can be opened, I'am sure SAR simulation approach will be a hot topic later on and it's my great honor to be the first group of researchers who benefit from this.

Cheers~

Florence

Dear Jingwen,

Thank you for your feedback.

Aspect 1 of yours was actually the core idea behind RaySAR. You can localize the 3D position of phase centers in world coordinates of the model and mark reflecting (model) surfaces contributing to selected salient image pixels (single to fivefold bounce with optional restriction to specular reflections). From my perspective, there are plenty of imaging effects left to explain as almost every object / building follows its own rule of signal composition :-).

As far as aspect 2 is concerned, both the amplitude and phase of signal contributions are provided by RaySAR. In the context of my work, the pixel phase was not of highest priority as I was mostly interested in the final phase center position in 3D space (comparable to the result of interferometric / tomographic processing). The value of the simulated pixel phase strongly depends on the degree of reality of the object model. 

Best regards

Stefan

Two days ago I was asked whether knowledge about the changes of the POV-Ray code (related to RaySAR) may be available. This information would be part of the package, of course (adapted C++ code and documentation where POV-Ray has changed). This effort should not be necessary to anybody again :-).

The next step of this project can be an open platform like OpenCV etc. which may serve for a greater educational or research purpose, allowing us to save time and contribute further. Specifically it will help me to identify and understand deterministic reflection effects occurring at individual man-made objects, buildings facade etc.

Finding objects in real life with which we can reach an adequate image localization accuracy equivalent to corner reflectors ? 

Thanks 

BR

Adnan Saeed

Thank you for your response. Indeed it will help you to understand deterministic effects you did not expect beforehand. For instance, understanding the nature of ghost persistent scatterers would not have been possible without the simulator. RaySAR is  like a door opener to reach the agreement of reason and reasoning.

The latter aspect you mention is a really hot issue. Indeed, we are following this direction at the moment. Therefore, a top-notch 3D model in absolute coordinates is the key element. The simulator remains the same all the time (for digital surface models as well as high-end 3D models).

Dear Stefan !

Did you came up to a decision of making RaySar available publicly ?

BR

Adnan

Yes, I will. It will just take some time to wrap everything to a package 😊.

I'd like to add one remark related to a question I received yesterday: whether RaySAR may be also applied to terrestrial applications.

Basically, RaySAR assumes objects to be related in the far field of the sensor where the signal wave front is expected to be quasi flat. This allows for the assumption of a locally constant signal incidence angle, which provides RaySAR with more flexibility in scene modeling and simulation. For instance, the far field assumption is okay for a small building block located hundreds of kilometers away from the SAR sensor in space.

For remarkable angle variations (more likely for terrestrial applications), the far field assumption may be inappropriate. Then, the orthographic projection in the range-elevation plane may be exchanged with a perspective one, what is not implemented so far. A possible workaround with the current RaySAR simulator may be to sub-sample the scene with several simulations where the corresponding signal incidence angle is varied. To this end, an appropriate sampling step width should be chosen for acceptable (small) discontinuities at the borders of neighboring images.

That is, RaySAR does not provide a straightaway solution but for sure provides a good basis for terrestrial tasks (using workarounds or adapted RaySAR solutions).

Okay, it's done :-). The RaySAR package can be downloaded on

https://github.com/StefanJAuer/RaySAR/releases

In there, the following components will be found:

- A readme to start with :-)

- A documentation of RaySAR (please read it carefully in advance of using RaySAR) and the methodological background (doctoral thesis)

- A license file related to the use of RaySAR (GNU v3)

- The RaySAR source code (additional POV-Ray code parts, MATLAB code) and a related guidance how to set up RaySAR

- Model files to test the functionality and properties of the RaySAR simulation environment

I'm looking forward to feedback from the community and new developments / research aspects based on RaySAR :-).

A dedicated Google Group for RaySAR improvements / applications can be found on

https://groups.google.com/forum/#!forum/raysar

Please take the opportunity to participate there in order to report your experiences and exchange ideas.

Good luck with your work.

By the way: I have created a topic in the Google Group

https://groups.google.com/forum/#!forum/raysar

where new RaySAR users are encouraged to post their first simulation results. In this context, I am curious to figure out which modeling environments are used and which workarounds may be necessary on the way to the goal.

Two months have passed now since the access to RaySAR was opened. I'm pretty surprised by the user interaction on the GitHub site so far and curious of the first simulation results to appear :-). In the meanwhile, I want to add a remark for users interested in SAR image simulation.

RaySAR was originally focused on 3D SAR applications (in the context of PSI and SAR Tomography) but can also be used to generate 2D SAR images. In that regard, interested users are also referred to a ISPRS paper where RaySAR is compared to two other state of the art image simulators based on ray tracing and rasterization techniques, see

https://www.researchgate.net/publication/270107408_Potentials_and_limitations_of_SAR_image_simulators_-_A_comparative_study_of_three_simulation_approaches

Article Potentials and Limitations of SAR Image Simulators - A Compa...

I would like to add one more remark on the RaySAR package which can be downloaded on the GitHub site.

The package includes a highly detailed 3D model of a building in Munich (absolute accuracy in the range of cm; model based on photogrammetry and geodetic field survey), which can be used to understand and evaluate the 3D simulation capability of RaySAR (note: the expected simulation results are included in the package, too).

Using the model files (povray model, dxf 3D model) and the attached simulation parameters, the results presented in

https://www.researchgate.net/publication/262232499_Linear_Signatures_in_Urban_SAR_Images_-_Partly_Misinterpreted?ev=prf_pub

can be confirmed and - if wanted - extended. In the paper at hand, the focus is on explaining the nature of building corner lines (often related to double reflection lines in the literature).

Article Linear Signatures in Urban SAR Images - Partly Misinterpreted?

FYI: The current status of RaySAR and aspects w.r.t. future applications are summarized in a paper at IGARSS; see https://www.researchgate.net/publication/305115067_RAYSAR_-_3D_SAR_SIMULATOR_NOW_OPEN_SOURCE

Conference Paper RaySAR - 3D SAR simulator: Now open source

FYI: RaySAR version 1.1 has been released on GitHub today, see:

https://github.com/StefanJAuer/RaySAR/releases

FYI: An updated RaySAR version 1.1 has been uploaded to GitHub (containing a hotfix to a message box bug). Sorry for the inconvenience.

FYI: the core elements of RaySAR (adapted ray tracer, image generation) can be integrated in automated procedures, see: 

https://www.researchgate.net/publication/312374400_Simulation-Based_Interpretation_and_Alignment_of_High-Resolution_Optical_and_SAR_Images

Article Simulation-Based Interpretation and Alignment of High-Resolu...

FYI: Here's an overview of publications reporting RaySAR applications or extensions since it has been made available.

Zhao, J. W., Wu, J. C., Ding, X. L., Zhang, L., and Hu, F. M.: "Analysis of multipath pixels in SAR images", Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLI-B7, 873-877, doi:10.5194/isprs-archives-XLI-B7-873-2016, 2016.

Link: http://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XLI-B7/873/2016/isprs-archives-XLI-B7-873-2016.pdf

Context: Analysis of dominant scatterers in SAR images

S. Auer, U. Balss, "Simulation-based Evaluation of Light Posts and Street Signs as 3-D Geolocation Targets in SAR Images", ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. IV-1/W1, 2017, pp. 12-18

Link: http://www.isprs-ann-photogramm-remote-sens-spatial-inf-sci.net/IV-1-W1/11/2017/isprs-annals-IV-1-W1-11-2017.pdf

Context: Analysis of light posts and street signs as prominent targets in SAR imaging

J. Guo, B. Lei, C. Ding and Y. Zhang, "Synthetic Aperture Radar Image Synthesis by Using Generative Adversarial Nets," in IEEE Geoscience and Remote Sensing Letters, vol. 14, no. 7, pp. 1111-1115, July 2017. doi: 10.1109/LGRS.2017.2699196

Link: http://ieeexplore.ieee.org/document/7927706/?source=tocalert&dld=dHVtLmRl

Context: Comparison with deep learning results

S. Auer, I. Hornig, M. Schmitt and P. Reinartz, "Simulation-Based Interpretation and Alignment of High-Resolution Optical and SAR Images," in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. PP, no. 99, pp. 1-15, doi: 10.1109/JSTARS.2017.2723082

Link: http://ieeexplore.ieee.org/document/7986985/

Context: Fusion of optical and SAR imagery in urban areas

S. Niu, X. Qiu, L. Peng and B. Lei, "Parameter Prediction Method of SAR Target Simulation Based on Convolutional Neural Networks," EUSAR 2018; 12th European Conference on Synthetic Aperture Radar, Aachen, Germany, 2018, pp. 1-5.

Link: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8437974&isnumber=8437291

Context: Definition of simulation parameters based on deep learning

David C. Mason, Sarah L. Dance, Sanita Vetra-Carvalho, Hannah L. Cloke, "Robust algorithm for detecting floodwater in urban areas using synthetic aperture radar images.," J. Appl. Remote Sens. 12(4), 045011 (2018), doi: 10.1117/1.JRS.12.045011.

Link: http://centaur.reading.ac.uk/80110/8/045011_1.pdf

Context: Simulation support for flood water detection

The papers may provide you with some orientation on how to make use of RaySAR in your field of work.