Control systems and methods for operating focal plane arrays are provided.
Digital imaging systems incorporating focal plane arrays having thousands or millions of photosensitive sites or pixels have become common in terrestrial cameras, and in cameras used in remote imaging applications, such as on satellites or autonomous vehicles. The focal plane arrays have the ability to operate in different modes. In particular, space-grade digital focal plane arrays (DFPAs) are highly configurable, offering a plethora of modes and settings. For example, they can be operated at different gain levels, and over different integration or exposure times. In addition, it is possible to operate focal plane arrays over different effective resolutions and using different areas of the array. These DFPAs are generally configured over a custom serial interface and usually require cycle-accurate synchronization with other clocks and/or control strobes provided to the sensor. Given this requirement, frame-to-frame and intra-frame commanding is not a supported concept of operations (CONOPS) in typical missions. In addition, control electronics have typically been incapable of enabling a full variety of imaging configurations that may be desirable using a contemporary focal plane array. Moreover, previously available control electronics have been incapable of maximizing temporal resolution of focal plane arrays.
In accordance with embodiments of the present disclosure, an imaging system incorporating reconfigurable focal plane electronics is provided. The system generally includes an image sensor, and in particular a digital focal plane array (DFPA). In addition, the system includes a microsequencer that interfaces with and controls operations of the DFPA. The system can also include or communicate with a host, from which commands regarding desired parameters for image acquisition operations are provided to the microsequencer, which in turn provides appropriate operational parameters or operations to the DFPA for execution of the image acquisition operations. In accordance with at least some embodiments, commands can be transmitted to the imaging system from a control authority that is located remotely relative to the imaging system. For example, the host system, DFPA and microsequencer can be located on a satellite or other vehicle, while the control authority can be located at a terrestrial location and can be in wireless communication with the imaging system.
In accordance with at least some embodiments of the present disclosure, the microsequencer is implemented by a reconfigurable focal plane electronics (RFPE) architecture. The microsequencer operates to command digital FPAs (DFPAs), including space-grade DFPAs that are highly configurable, and that offer a plethora of modes and settings. A microsequencer implemented by the reconfigurable focal plane electronics in accordance with embodiments of the present disclosure provides a high-speed interface and sophisticated control logic that allows the DFPA settings to be changed on a frame-to-frame or even intra-frame basis, allowing for dynamic reconfiguration of the DFPA and complex imaging sequences including high-dynamic range modes, low noise sample-up the ramp, or staggering window collects.
In accordance with further embodiments of the present disclosure, the microsequencer implements a serial interface providing cycle-accurate synchronization with other clocks and/or control strobes provided to the sensor. Moreover, frame-to-frame and intra-frame commanding is supported. The microsequencer can be implemented using RFPEs that include a field programmable gate array (FPGA), and can provide the real-time, frame-to-frame commanding capability in an integrated fashion.
Methods in accordance with embodiments of the present disclosure include providing a DFPA and microsequencer. The microsequencer contains a plurality of commands that can be applied to the DFPA. In response to receiving imaging task commands from a host related to the operation of the DFPA, the microsequencer operates to determine a command, set of commands, or sequence of commands that should be selected for implementing the task. The commands selected by the microsequencer can be stored locally, and can contain particular registers and values appropriate for the associated DFPA. Commands are formatted and passed to the DFPA through operation of a state machine.
Additional features and advantages of embodiments of the present disclosure will become more readily apparent from the following description, particularly when taken together with the accompanying drawings.
Although the platform 100 depicted in the figure is shown as a satellite, a platform 100 can include other types of vehicles or systems, such as but not limited to space craft, airplanes, unmanned aircraft, balloons, aerostats, terrestrial vehicles, or static platforms. In accordance with at least some embodiments of the present disclosure, the imaging system 104 is operated in association with a control authority 120 that is in communication with the imaging system over a communication link 124, but that is located remotely from the imaging system 104. For example the control authority 120 can be an earth based system in communication with the imaging system 104 over a wireless communication link 124. In accordance with embodiments of the present disclosure, the control authority 120 can send control signals to the imaging system 104, automatically or in response to a command from a user, and can receive image data and associated status telemetry data from the imaging system.
As an example, but without limitation, the digital focal plane array 204 can include a backside illuminated CMOS image sensor having a 1024×1024 two-dimensional array of pixels. As can be appreciated by one of skill in the art after consideration of the present disclosure, in addition to a focal plane array formed from a plurality of photosensitive sites or pixels, the focal plane array 204 can incorporate or be associated with driver and analog-to-digital conversion (ADC) circuitry, enabling the focal plane array 204 to provide a digital output representative of an amplitude or intensity of light detected at each pixel within the focal plane array 204.
The RFPE 208 can include memory 224, data storage 228, and a processor 232. The RFPE memory 224 can include volatile or non-volatile solid-state memory, such as DRAM, SDRAM, or the like. The RFPE memory 224 can provide short or long term storage for instructions 236 that are executed by the RFPE processor 232 to implement operations or functions of the DFPA 204 and the RFPE 208 and imaging system 104 more generally. The RFPE memory 224 can also store configurable parameters, image commands, including image command that are particular to the DFPA 204, intermediate data products, output data, and the like. The RFPE memory 224 can additionally provide a first-in first-out memory function. In accordance with further embodiments of the present disclosure, the application instructions 236 can include instructions that implement the microsequencer 212. The RFPE data storage 228 generally includes non-volatile data storage, such as flash memory, solid-state drives, hard disk drives, optical disk drives, erasable programmable ROM, and the like. The RFPE data storage 228 can provide short or long term storage for application instructions 236 that are executed by the RFPE processor 232, configurable parameters, intermediate data products, output data, and the like. In accordance with still other embodiments of the present disclosure, the RFPE data storage 228 can store a library of image commands or frame commands for controlling operation of the DFPA 204.
The RFPE processor 232 can include one or more field-programmable gate arrays (FPGAs), multi-threaded processors, graphics processing units (GPUs), general purpose processors, or the like. For example, the RFPE processor 232 can be formed from an FPGA processor, alone in combination with multiple GPUs or other devices. In accordance with further embodiments of the present disclosure, the RFPE 208 can include a plurality of boards or cards, with each board including memory 224 and a GPU or other processor 232.
The host system 216 can include a processor 238, memory 240, data storage 244, and a communications interface 248. The processor 238 can include a general purpose programmable processor or the like. The memory 240 can include, for example, volatile or non-volatile memory, and can provide short or long-term storage for application programming or instructions, control parameters, intermediate data products, data, or the like. The data storage 244 generally includes non-volatile storage for application programming or instructions, control parameters, intermediate data products, data, or the like. The communications interface 248 can support the passing of commands to and the receiving of data from the RFPE 212. In addition, the communications interface 248 can support communications between the imaging system 104 and remote systems, control authorities 120, or communication nodes.
In operation, the host system 216 sends commands to the RFPE 208. The commands can be stored locally by the host system 216, can be transmitted to the host system 216 from a control authority 120, and/or can be generated by the host system 216. In accordance with embodiments of the present disclosure, the commands can include imaging tasks that define aspects of imaging operations to be performed by the imaging system 104. The RFPE 208 itself, or through a microsequencer 212 function implemented by the RFPE 208, in turn provides instructions to the digital focal plane array 204 that configure the digital focal plane array 204 appropriately for completing imaging tasks, and that operates the digital focal plane array 204 to obtain an image in accordance with the imaging tasks or other instructions received from the host system 216 or a control authority 120. Moreover, the microsequencer 212 can control the digital focal plane array 204 to obtain a sequence of images in response to a command received from the host system 216. The host system 216 can also perform functions related to operation of the platform 100 and/or the imaging system 104, including but not limited to: operations relating to the positioning of the platform 100; receiving and acting on instructions from a command center 120; transmitting images collected by the imaging system 104; transmitting the results of image analysis performed by the RFPE 208; and performing actions, including but not limited to actions concerning the positioning of the platform 100, and actions taken by the platform 100 relative to objects 116 within a scene.
With reference now to
The image collected according to the first image command is read out from the focal plane array 204 to the microsequencer 212 (step 312). In accordance with embodiments of the present disclosure, the image frames are each associated with an execution time tag, enabling different images to be precisely synchronized with platform 100 system time or host payload time. Alternatively, commands can be executed by the microsequencer 212 autonomously. The microsequencer 212 then configures the focal plane array 204 for a second or next image (step 316). The next image can be taken as a result of the execution of an image command selected as complying with the same imaging task that resulted in the selection of the first image command, or it can be selected in response to another imaging task. The configuration of the focal plane array 204 for the second image can be the same as or different from the configuration used for the first or previous image. Accordingly, the inter-frame configuration of a sequence of frames can be changed. In accordance with still further embodiments, the settings can be changed during an integration period and/or while the commands for a frame are being executed. In addition, the image data obtained by the focal plane array is transmitted back to the host (step 320). This process can continue, with each subsequent image being the same as or different from the preceding image.
The imaging commands 406 can include predefined imaging commands 406 that can be selected for inclusion in a command sequence 405 by a user or in response to a task received from a user from a library or store of such commands 416 maintained in memory 224 or data storage 228 included as part of the reconfigurable focal plane electronics 208, stored in memory 240 or data storage 244 included in the host system 216, or stored in an external memory or data storage. Alternatively or in addition, imaging commands 406 can be customized or otherwise created by a user or control authority 120, either through modifications to predefined commands 406, or through the creation of entirely new commands. Each image command 406 generally includes DFPA 204 operating parameters that are applied in connection with acquiring a frame or a set of frames of image data by the associated DFPA 204. In accordance with embodiments of the present disclosure, the image commands 406 included in a particular instance of a microsequencer 212 are specific to the DFPA 204 with which the microsequencer is deployed. Accordingly, different sets of frame commands 406 are typically associated with different DFPAs 204.
The frame commands 406 can include one or more fields that are each associated with a least one DFPA 204 operating parameter and an associated value. In accordance with at least some embodiments of the present disclosure, a stack of baseline register values 412 that contain fields of configuration parameters 440 and corresponding values 444 is maintained in the memory 224 or data storage 228 of the RFPE 208. In such embodiments, each image command 406 only includes a field for a particular parameter 440 in which the commanded value 444 is different than a baseline or default value for that parameter. Moreover, the given value for each parameter can be a difference or deviation from the baseline register value, referred to herein as a difference value 448. Accordingly, execution of the individual image definitions stored as part of image commands 406 can be made with reference to the baseline register values. Alternatively, the individual frame definitions can be set forth in absolute terms, in which case the baseline values need not be stored.
Each command sequence 405 can be generated or established in response to user commands 436 that specify desired parameters for an imaging task that includes the acquisition of one or more images, but that are not themselves capable of operating the DFPA 204. The commands 406 suitable for implementing the operations directed by a particular user command 436 regarding an imaging task can be selected by operation of the microsequencer 212. The set of commands 406 forming the command sequences 405 are placed in a command queue 408 established locally in memory 224 or data storage 228. From the command queue 408, a set of image commands 406 included in a first command sequence 405a is loaded into the state machine 404. In accordance with embodiments of the present disclosure, only the commands 406 included in a single command sequence 405 are loaded into the state machine 404 at any one point in time. In such embodiments, any other command sequences 405 remain in the command queue 408. For example, as shown in the figure, a second command sequence 405b and an nth command sequence 405n are maintained in the command queue 408 while the first command sequence 405a is loaded into the state machine 404. Alternatively, commands 406 in series from a command queue 406. Any number of specific command sequences 405, containing any number of frame definitions 406, can be stored, subject to the limitations of the memory 224 and data storage 228. In addition, specific command sequences 405 can be repeated or arranged in any order.
The state machine 404 can be operated in connection with control and status commands 410 that are passed between the host system 216 and the RFPE 208 implementing the state machine 404. The control and status commands 410 can include commands to initiate or stop the execution of frame commands. When operation is initiated, the state machine 404 loads the next command sequence 405 in the command queue 408, here sequence 405a, shown already loaded in the state machine 404, and cycles through the image commands or definitions 406. In this example, the first command sequence 405a includes five frame commands 406. Additional command sequences 405, i.e. any command sequences 405 placed in the RFPE 208 but not currently being executed by the state machine 404, remain in the queue 408. The image commands 406 included within a particular command sequence 405 are executed using timing information provided by a system clock or time source 420. In addition, the state machine 404 supplies a focal plane array command stream 424 consisting of the frame parameters according to supplied system timing information 420 to a focal plane array command formatter 428. The system timing information can be provided by the host system 216. The focal plane array command formatter 428, in accordance with the least some embodiments of the present disclosure, is implemented through application programming executed by the microsequencer 212. Alternatively, the focal plane array command formatter 428 can be implemented separately, for example as a field programmable gate array or controller provided separately from other components of the microsequencer 212. The focal plane array command formatter 428 provides a DFPA clock signal 432 and DFPA data 436, in the form of the parameter information, to the focal plane array 204, to thereby control operation of the focal plane array 204.
Accordingly, an RFPE 208 as disclosed herein provides cycle-accurate synchronization of image acquisition and DFPA 204 operation with system clocks and control strobes. More particularly, the microsequencer 212 functions and capabilities enable real-time, frame-to-frame commanding of the associated DFPA 204. The image commands 406 maintained locally in the command store 416 further allow a control authority 120 or user to select or command image operations, without requiring that detailed control parameters be uploaded to the RFPE 208. Moreover, the image commands 406 that are stored locally in the command store 416 can include commands that involve the acquisition of multiple image frames, having the same or different image parameters. Accordingly, aspects of the control of the associated DFPA 204 are abstracted by the RFPE 208.
Image commands 406 for carrying out the defined imaging task are then selected by the RFPE 208 (step 508). For example, this step may be performed by the execution of application programming 236 by the processor 232 of the RFPE 208. Selecting image commands 406 can include selecting one or more image commands 406 for inclusion in a sequence 405 of image commands 406. In addition, time tags 414 are assigned to the individual image commands 406 (step 512). For example, where the imaging task includes acquiring a sequence of images as the platform moves relative to the surface of the earth 112 or another body, the time tags 414 can be assigned to ensure that a desired overlap between adjacent images, a desired area of the body, or the like is achieved. The time tags 414 can be in terms of an absolute reference time, a system time, or a relative time. The image commands 406 in a sequence 405 can express a sequence of frame parameters or commands for execution by the RFPE 208, in order to obtain a set or sequence of images according to parameters specified by the related imaging task. In addition, the image commands 406 can be predefined and stored as part of the RFPE 208, on the host system 216, or on a control authority 120 system, can be acquired from some other source, or can be built by a control authority 120 or user. Each image command 406 can include the acquisition of one or more images or frames of data. For example, an image command 406 directed to the acquisition of a high dynamic range image can include the acquisition of multiple frames of image data in quick succession at different gain levels and/or different exposure or integration times. The image command or commands 406 selected and time tagged in connection with the imaging task are then assigned to a command sequence 405, and placed in the command queue 408 (step 516). Alternatively, the image command or commands 406 and associated time tags 414 can be placed in the command queue 408, without being explicitly assigned to any command sequence 405. In accordance with still other embodiments, entire command sequences 405 can be predefined and selected as appropriate in response to particular imaging tasks. At step 520, a determination can be made as to whether another imaging task remains to be handled. If another imaging task needs to be handled, the process can return to step 508, and commands 406 for the tasks can be selected by the RFPE 208.
If all received imaging tasks have been handled a determination can be made as to whether a control signal 410 to start execution of image commands 406 held in the command queue 408 has been received (step 524). If no such command has been received, the process can return to step 520 to determine whether a new imaging task has been received. In response to receiving a control signal 410 to start execution of imaging tasks, a command sequence 405 or set of image commands 406 and associated time tags 414 are loaded into the state machine 404 (step 528). Loading the image commands 406 into the state machine 404 can include the microsequencer 212 beginning a read operation at a first or next memory address within the command queue 408 containing a command 406.
The state machine 404 creates a DFPA 204 command for each image command, and stores the created command stream in a first-in first-out memory included in the RFPE 208 (step 532). The state machine 404 then commands an output serializer implemented by the microsequencer 212 to send a command stream 424 containing the contents of the first-in first-out memory to the digital focal plane array 204. The command for the current image command 406 or frame can be provided to the DFPA 204 by a focal plane array command formatter 420. The state machine 404 sends the commands 406 to the DFPA 204 for execution at the times indicated by the associated time tags 414 (step 536). For example, after sending the command stream 424 related to the first image command 406, the state machine 404 waits for a period of delay, defined by the specific image command 406. After the period of delay has elapsed, the state machine 404 sends parameters of the next image command 406 as a DFPA control stream 424. In the example illustrated in
After all of the image commands 406 have been executed or after the execution of the commands 406 in the current command sequence 405 (e.g. sequence 405a) (step 540), the state machine 404 determines whether additional image commands 406 or command sequences 405 are waiting in the command queue (step 544). If an additional image command 406 or command sequence 405 (e.g. sequence 405b) is waiting in the command queue 408, the command or commands 406 are loaded into the state machine 404 and are executed (i.e. the process returns to step 528).
After all of the command sequences 405 in the command queue 408 have been loaded into the state machine 404 and executed, a determination can be made as to whether a command to re-execute some or all of the command sequences 405 held in the command queue has been received or is otherwise in place (step 548). If re-execution is indicated, the process can return to step 516. If re-execution is not indicated, a determination can be made as to whether other imaging tasks have been received (step 552). If additional imaging tasks have been received, the process can return to step 508. Otherwise the process can end.
With reference now to
An imaging system 104 as discussed herein can provide the following capabilities and advantages over other systems:
1) Defining a sequence of time commands, such as combining multiple image collects, typically with the DFPA in a unique gain mode, setting of window modes, or a unique set of integration period during each image, to form a single composite image with extended dynamic range. Other examples include co-adding multiple images taken with identical settings (to improve SNR) and summing hyperspectral line imagery to form 2D spatial images.
2) Implementing a temporal component that provides dynamic, frame-by-frame and intraframe control of the DFPA in different imaging modes, providing each command with an associated execution time tag, so image collections can be precisely synchronized with platform (e.g., spacecraft) system time enabling new mission CONOPS, and
3) Providing an abstraction layer for the DFPA interface to enable a set of common imaging modes (including both simple and sophisticated sequences) that can be reused for different missions. In addition, new image sequences can be uploaded to the RFPE while on orbit, thereby enabling the RFPE 208 to be reconfigured for different missions or to enable new CONOPS over the full life of the instrument.
The foregoing discussion has been presented for purposes of illustration and description. Further, the description is not intended to limit the disclosed systems and methods to the forms disclosed herein. Consequently, variations and modifications commensurate with the above teachings, within the skill or knowledge of the relevant art, are within the scope of the present disclosure. The embodiments described hereinabove are further intended to explain the best mode presently known of practicing the disclosed systems and methods, and to enable others skilled in the art to utilize the disclosed systems and methods in such or in other embodiments and with various modifications required by the particular application or use. It is intended that the appended claims be construed to include alternative embodiments to the extent permitted by the prior art.
This application claims the benefit of U.S. Provisional Patent Application Ser. No. 62/812,595, filed Mar. 1, 2019, the entire disclosure of which is hereby incorporated herein by reference.
Number | Name | Date | Kind |
---|---|---|---|
5588073 | Lee et al. | Dec 1996 | A |
5719794 | Altshuler et al. | Feb 1998 | A |
5960391 | Tateishi et al. | Sep 1999 | A |
6075991 | Raleigh et al. | Jun 2000 | A |
6252627 | Frame et al. | Jun 2001 | B1 |
6437692 | Petite et al. | Aug 2002 | B1 |
6597394 | Duncan et al. | Jul 2003 | B1 |
6820053 | Ruwisch | Nov 2004 | B1 |
7020501 | Elliott et al. | Mar 2006 | B1 |
7590098 | Ganesh | Sep 2009 | B2 |
8019544 | Needelman et al. | Sep 2011 | B2 |
8583371 | Goodzeit et al. | Nov 2013 | B1 |
8929936 | Mody et al. | Jan 2015 | B2 |
9073648 | Tsao et al. | Jul 2015 | B2 |
9191587 | Wright et al. | Nov 2015 | B2 |
9294365 | Misra et al. | Mar 2016 | B2 |
9449374 | Nash et al. | Sep 2016 | B2 |
9702702 | Lane et al. | Jul 2017 | B1 |
9924522 | Gulati et al. | Mar 2018 | B2 |
9927510 | Waldron et al. | Mar 2018 | B2 |
10021313 | Chen et al. | Jul 2018 | B1 |
10048084 | Laine et al. | Aug 2018 | B2 |
10271179 | Shima | Apr 2019 | B1 |
10970520 | Kim et al. | Apr 2021 | B1 |
20050049876 | Agranat | Mar 2005 | A1 |
20050160406 | Duncan | Jul 2005 | A1 |
20050228660 | Schweng | Oct 2005 | A1 |
20060030332 | Carrott et al. | Feb 2006 | A1 |
20070010956 | Nerguizian et al. | Jan 2007 | A1 |
20080020354 | Goree et al. | Jan 2008 | A1 |
20080045235 | Kennedy et al. | Feb 2008 | A1 |
20080293353 | Mody et al. | Nov 2008 | A1 |
20090179142 | Duparre et al. | Jul 2009 | A1 |
20090197550 | Huttunen et al. | Aug 2009 | A1 |
20090268619 | Dain et al. | Oct 2009 | A1 |
20100091017 | Kmiecik et al. | Apr 2010 | A1 |
20120071105 | Walker et al. | Mar 2012 | A1 |
20120072986 | Livsics et al. | Mar 2012 | A1 |
20120163355 | Heo et al. | Jun 2012 | A1 |
20120167144 | Avison-Fell | Jun 2012 | A1 |
20120202510 | Singh | Aug 2012 | A1 |
20120238201 | Du et al. | Sep 2012 | A1 |
20120238220 | Du et al. | Sep 2012 | A1 |
20140218520 | Teich et al. | Aug 2014 | A1 |
20140232871 | Kriel et al. | Aug 2014 | A1 |
20140282783 | Totten et al. | Sep 2014 | A1 |
20140329540 | Duggan et al. | Nov 2014 | A1 |
20150009072 | Nijsure | Jan 2015 | A1 |
20160101779 | Katoh | Apr 2016 | A1 |
20160173241 | Goodson et al. | Jun 2016 | A1 |
20160187477 | Wang | Jun 2016 | A1 |
20170120906 | Penilla et al. | May 2017 | A1 |
20170123429 | Levinson et al. | May 2017 | A1 |
20170366264 | Riesing et al. | Dec 2017 | A1 |
20180019910 | Tsagkaris et al. | Jan 2018 | A1 |
20180025641 | LaVelle et al. | Jan 2018 | A1 |
20180033449 | Theverapperuma et al. | Feb 2018 | A1 |
20180053108 | Olabiyi et al. | Feb 2018 | A1 |
20180082438 | Simon et al. | Mar 2018 | A1 |
20180107215 | Djuric et al. | Apr 2018 | A1 |
20180149730 | Li et al. | May 2018 | A1 |
20180268571 | Park et al. | Sep 2018 | A1 |
20180293893 | Yang et al. | Oct 2018 | A1 |
20180324595 | Shima | Nov 2018 | A1 |
20190049955 | Yabuuchi et al. | Feb 2019 | A1 |
20190066713 | Mesgarani et al. | Feb 2019 | A1 |
20190122689 | Jain et al. | Apr 2019 | A1 |
20190164430 | Nix | May 2019 | A1 |
20190213887 | Kitayama et al. | Jul 2019 | A1 |
20190294108 | Ozcan et al. | Sep 2019 | A1 |
20190318725 | Le Roux et al. | Oct 2019 | A1 |
20190322282 | Theodosis et al. | Oct 2019 | A1 |
20190353741 | Bolster, Jr. et al. | Nov 2019 | A1 |
20190363430 | Wang et al. | Nov 2019 | A1 |
Number | Date | Country |
---|---|---|
108875595 | Nov 2018 | CN |
Entry |
---|
U.S. Appl. No. 16/597,411, filed Oct. 9, 2019, Schmidt et al. |
U.S. Appl. No. 16/668,826, filed Oct. 30, 2019, Shima. |
U.S. Appl. No. 16/693,992, filed Nov. 25, 2019, Tchilian. |
U.S. Appl. No. 16/745,725, filed Jan. 17, 2020, Tchilian et al. |
“Deep Learning Meets DSP: OFDM Signal Detection,” KickView Tech Blog, Feb. 13, 2018, 25 pages [retrieved online from: blog.kickview.com/deep-learning-meets-dsp-ofdm-signal-detection/]. |
Buchheim “Astronomical Discoveries You Can Make, Too!” Springer, 2015, pp. 442-443. |
Ma et al. “Attitude-correlated frames approach for a star sensor to improve attitude accuracy under highly dynamic conditions,” Applied Optics, Sep. 2015, vol. 54, No. 25, pp. 7559-7566. |
Ma et al. “Performance Analysis of the Attitude-correlated Frames Approach for Star Sensors,” IEEE, 3rd IEEE International Workshop on Metrology for Aerospace (MetroAeroSpace), Firenze, Italy, Jun. 22-23, 2016, pp. 81-86. |
Nair et al. “Accelerating Capsule Networks with Tensor Comprehensions,” Princeton, May 2018, 8 pages. |
Ni et al. “Attitude-correlated frames adding approach to improve signal-to-noise ratio of star image for star tracker,” Optics Express, May 2019, vol. 27, No. 11, pp. 15548-15564. |
Wang “Research on Pruning Convolutional Neural Network, Autoencoder and Capsule Network,” before Oct. 9, 2018, 11 pages. |
Wang et al. “An Optimization View on Dynamic Routing Between Capsules,” ICLR 2018 Workshop, Feb. 2018, 4 pages. |
Number | Date | Country | |
---|---|---|---|
62812595 | Mar 2019 | US |