Imaging Subsystem
Imaging Calculations and Technical Details
| lensimagesensingsystemofcube_final_(1).pdf |
Camera and Imaging System
Using a modified version of the GomSpace NanoCam C1U camera and integrated imaging system, we are able to achieve a 10 meter per pixel spatial resolution.
The original C1U had a pixel ratio of 2048 x 1536 and a Field of View of 9.22 degrees, and an integrated processor and 2GB flash storage device for processing, compressing, and storing the images. However, with that pixel resolution and lens FoV, the camera was only capable of achieving a spatial resolution of about 47 meters per pixel at the altitude of 450 km, far too coarse to resolve individual buildings and features on the ground.
To get around this and achieve the goal of 10 meters per pixel, CubeEyes contracted with GomSpace to produce an improved camera module based on the original C1U. This new system uses a larger imaging sensor with a pixel resolution of 6576 x 4384, and a custom lens from Schneider Optics with a longer focal length to get an improved, narrower field of view of 8.36 degrees. However, using this improved system cost more space within the CubeSat, and bumped the design from a 1U (1o00 cm^3) to a longer 3U (3000 cm^3) satellite body.
Original NanoCam C1U
http://gomspace.com/?p=products-c1u
Custom Order Lens From Schneider Optics
https://www.schneideroptics.com/industrial/custom_solutions/custom.htm
OnSemi 28 Megapixel Sensor
http://www.onsemi.com/pub_link/Collateral/KAI-29050-D.PDF
Data Requirements
At 450 km, the orbital period for each satellite will be 90 minutes. In order for each longitudinal region (calculations based on rotation at the equator as a worst case estimate) to be photographed as the planet rotates, a satellite will need to pass over the equator every 141.34 seconds, giving a minimum satellite count of 38. As the satellite flies over the sunlit side of the Earth, it will take roughly 457 images. Given that a 6576 x 4384 image full of RGB noise can be compressed to about 3.8 MB (through internal empirical testing, we found that a JPEG compressor program set to "10% quality" can still show enough detail), that is 1737.3 MB of data per satellite, far too much for a single satellite to transmit over the course of one orbit of the planet given a 2 Mbps downlink. In order to cut this burden in half, to a more manageable maximum of 868.65 MB of data, the constellation was doubled, plus 4 for contingency. Each satellite will acquire an image every 11.8 seconds.
Additionally, if a customer is not interested in detailed images of the ocean, the satellites can be configured to use higher compression settings when the GPS system calculates that it is over water. These reduced-quality images can be as small as 331 KB, versus the typical empirical value we found as 3.8 MB. This can result in a brief reduction in operating costs as fewer satellites will be needed to collect data in this mode.
Using a modified version of the GomSpace NanoCam C1U camera and integrated imaging system, we are able to achieve a 10 meter per pixel spatial resolution.
The original C1U had a pixel ratio of 2048 x 1536 and a Field of View of 9.22 degrees, and an integrated processor and 2GB flash storage device for processing, compressing, and storing the images. However, with that pixel resolution and lens FoV, the camera was only capable of achieving a spatial resolution of about 47 meters per pixel at the altitude of 450 km, far too coarse to resolve individual buildings and features on the ground.
To get around this and achieve the goal of 10 meters per pixel, CubeEyes contracted with GomSpace to produce an improved camera module based on the original C1U. This new system uses a larger imaging sensor with a pixel resolution of 6576 x 4384, and a custom lens from Schneider Optics with a longer focal length to get an improved, narrower field of view of 8.36 degrees. However, using this improved system cost more space within the CubeSat, and bumped the design from a 1U (1o00 cm^3) to a longer 3U (3000 cm^3) satellite body.
Original NanoCam C1U
http://gomspace.com/?p=products-c1u
Custom Order Lens From Schneider Optics
https://www.schneideroptics.com/industrial/custom_solutions/custom.htm
OnSemi 28 Megapixel Sensor
http://www.onsemi.com/pub_link/Collateral/KAI-29050-D.PDF
Data Requirements
At 450 km, the orbital period for each satellite will be 90 minutes. In order for each longitudinal region (calculations based on rotation at the equator as a worst case estimate) to be photographed as the planet rotates, a satellite will need to pass over the equator every 141.34 seconds, giving a minimum satellite count of 38. As the satellite flies over the sunlit side of the Earth, it will take roughly 457 images. Given that a 6576 x 4384 image full of RGB noise can be compressed to about 3.8 MB (through internal empirical testing, we found that a JPEG compressor program set to "10% quality" can still show enough detail), that is 1737.3 MB of data per satellite, far too much for a single satellite to transmit over the course of one orbit of the planet given a 2 Mbps downlink. In order to cut this burden in half, to a more manageable maximum of 868.65 MB of data, the constellation was doubled, plus 4 for contingency. Each satellite will acquire an image every 11.8 seconds.
Additionally, if a customer is not interested in detailed images of the ocean, the satellites can be configured to use higher compression settings when the GPS system calculates that it is over water. These reduced-quality images can be as small as 331 KB, versus the typical empirical value we found as 3.8 MB. This can result in a brief reduction in operating costs as fewer satellites will be needed to collect data in this mode.