Radiometric calibration and optical analysis of in-flight satellite-based imagers is generally achieved using on-board systems and by vicarious observation of celestial objects and earth-based geographical locations. These methods have been the standard for most instruments currently deployed; however, on-board systems are costly and existing vicarious means have known drawbacks which impact uncertainty in final measurements.
Algorithms developed by RIT in collaboration with the National Gallery of Art, and their associated reference data sets, have been optimized using reflectance image cubes collected from early Italian renaissance illuminated manuscripts. To date, results from applying these optimized algorithms on other early Italian renaissance illuminated manuscripts has shown great promise. The next step is to further test and develop the algorithms using reflectance image cubes acquired from paintings created by the same or similar artist from the same time period.
This project performs fundamental research into the end-to-end performance and evaluation of remote sensing imaging systems and the exploitation of image data from such systems. The overall focus is to perform research on topics such as: 1) semantic understanding of point cloud data (i.e., 3D /LiDAR processing), 2) novel multi-modal image registration techniques (using CNN’s), 3) anomaly, change and glint-type detection algorithms, 4) shadow detection in aerial imagery, 5) UAS data collections, processing, and storage techniques, 6) production of pixel-wise confidence estimates in disparity maps produced through stereoscopic techniques, and 7) DIRSIG SAR capabilities and image generation.
This project is aimed at developing an automated software system to ingest commercial satellite imagery and automatically detect the presence of certain objects or changes. RIT has been supporting this project with several tasks. 1) Image Chip Simulation – We have been using DIRSIG to simulate commercial satellite imagery of various objects for use in training deep learning algorithms to perform object detection. 2) Multispectral Change Detection Algorithm Evaluation and Development – We developed a Python toolbox for change detection evaluation and implemented a number of state-of-the-art algorithms for evaluation. The figure below presents an example change detection pair and a screen shot of the toolbox interface. 3) Unstructured Change Detection with Class Agnostic Region Proposals – We investigated the use of region proposals to detect unstructured changes in image pairs. 4) Unstructured Change Detection with LambdaNet – We also investigated the use of the LambdaNet neural network architecture to detect unstructured changes in image pairs.
The centerpiece of this project is a state-of-the-art integrated mast-mounted hyperspectral imaging system developed by Dr. Bachmann, his students, and RIT staff. The system includes the Headwall visible and near-infrared (VNIR) E-Series micro-Hyperspec High-Efficiency (micro-HE) imager, a General Dynamics maritime-rated high-speed pan-tilt unit, and an onboard Vectornav GPS-IMU system for pointing, georeferencing, and precision timestamps. At maximum operating rates, the system can produce a low-rate hyperspectral video imagery time series at about 1.5 Hz. Details of the system and purpose are further described in a recent journal article that was featured on the cover of the Journal of Imaging.
As NASA prepares to define the next generation of Landsat earth observing systems, this project has been exploring requirements for these future systems. Topics studied include: a) Impacts of wide-swath imaging including the potential for spectral shifts in channel bandpasses; b) Improved vegetation parameter retrievals from additional red-edge spectral bands; c) Methods for characterization and specification of stray light requirements in reflective spectral bands; d) Requirements for temporal sampling and satellite re-visit rates in agricultural applications; e) Sensitivity of spectral band artifacts in longwave infrared band filters to surface temperature retrieval accuracy; f) Efficient and realistic simulation of vegetation growth and resulting imagery.
This project focuses on using spectral imagery to retrieve concentrations of waterbody components using modeled Look-Up-Tables (LUTs) to determine how imaging systems could be improved for this task, e.g., Landsat. This work extends the LUT retrieval method to assess its ability to retrieve pigments related to harmful cyanobacteria blooms. Imagery from Landsat satellites, as well as multi and hyperspectral unmanned aerial system (UAS) is used for this assessment.
Prescribed fire is used as a management tool for a number of reasons- fuel reduction for mitigating the impact of wild fires, habitat improvement for game and other species, and as a successional management tool for forests like the coastal plains barrens, among others. Current predictive models used to simulate fire behavior during low-intensity prescribed fires (and low-intensity wildfires) are empirically-based, simplistic, and fail to adequately capture variability in fuel characteristics and interactions with important meteorological variables. We are undertaking, along with collaborators from the USDA Forest Service - Northern Research Station, -Pacific Northwest Research Station, University of Edinburgh and the Worcester Polytechnic Institute, to use a suite of measurements at the fuel particle, fuel bed, field plot and stand scales (or levels) to quantify how variability in fuel characteristics and key meteorological factors interact to drive fire behavior during low intensity prescribed burns.
Advancements permitting the rapid extraction of 3D point clouds from stereo imagery covering large portions of the landscape have provided a vast collection of high-fidelity digital surface models of the planetary surface. These models should be acquired rapidly in time, presenting an opportunity for surface model change detection on a quasi-global scale. Although two-dimensional change detection from remotely sensed imagery is fairly common, leveraging 3D data derived from stereo pairs present new challenges.
The Fulbright Specialist Program is a grant to a U.S. domain expert to participate in a project proposed by the host country. Professor Vodacek was selected for a project in Malaysia proposed by Prof. Tee Yee Kai of Universiti Tunku Abdul Rahman (UTAR). The project concerns extraction of sound from high speed video in a process pioneered at MIT.
Several Split Window algorithms have been successfully applied to infrared satellite data (e.g. MODIS, VIIRS) for the estimation of land surface temperature. This research focuses on creating a Split Window algorithm for the Thermal Infrared Sensor (TIRS) onboard Landsat 8. Results are compared to the current Single Channel surface temperature method and 1518 in-situ validation sites. These sites include the SURFRAD and Ameriflux networks as well as ocean buoy measurements (NOAA) and Lake Tahoe and Salton sea buoys (JPL). Results of this study are provided to the Unit- ed States Geological Survey (USGS) to support their goals of releasing a validated land-surface temperature product to users.
This project is supporting the development of an airborne multiband infrared imaging instrument using an uncooled microbolometer array under funding from NASA’s Instrument Incubator Program (IIP). During AY 2018-19 efforts continued on studying the ability of the multiband instrument to detect enhanced levels of methane gas near the surface. This effort was pursued through the study of existing airborne longwave infrared imaging spectrometer data as well as through radiative transfer modeling using MODTRAN6. The figure to the right shows example results of the modeling study. This figure shows the sensitivity study of applying a detection algorithm to simulated data for the DRS multispectral instrument (MURI).
The objective of this research is to assess the feasibility of using, in order of priority, remotely captured thermal infrared, LiDAR, hyperspectral, and RGB imagery to detect and classify duck nests in the Pothole Prairie region, North Dakota. Natural resources such as ducks and wildlife are vital to North Dakota, bringing in more than 21 million visitors and 3 billion dollars in revenue annually. 50-80% of all North American ducks are born in the Prairie Pothole Region which ranges from Canada down through Montana, the Dakotas, Minnesota, and finally, Iowa. The Prairie Pothole Region is shrinking, losing approximately 50,000 acres every year. It is important that we can accurately assess the duck populations as well as the environment they thrive in so that we can understand the impacts of industrial development, conservation efforts, and population management techniques.
Assessing crop health and, moreover, predicting crop yield from remote sensing imagery is an ongoing area of interest in precision agriculture. We thus pursue a unified spectral-structural model of the growth and development of corn (Zea mays L.) which considers spatiotemporally-varying environmental variables, in pursuit of a theoretically optimal satellite system specification through light-transport simulation with DIRSIG5 (to the right). Our preliminary objective is to assess the fidelity of a virtual corn field in accurately reproducing spectral response of a spaceborne imaging platform. Our overarching objective is to leverage such a simulation-based approach to perform a spaceborne system-level analysis to establish minimum requirements for assessing corn spectral-structural variability and associated yield-modeling throughout the growing season. The outcomes associated with these objectives could contribute to a definitive specification for optimized corn yield monitoring.
RIT has been a major participant in calibrating Landsat’s thermal archive for over twenty years. In this work, a forward modeling process has been developed and refined to make accurate estimates of band-effective radiance for the thermal instruments onboard Landsat 4, 5, 7, & 8. This model ingests temperature measurements obtained from the National Oceanic and Atmospheric Administration (NOAA) buoys as well as local radiosonde data to describe the environmental conditions of the pixel of interest during a Landsat overpass.
The Landsat program is planning to release a Land Surface Temperature (LST) product as part of the Landsat Collection 2 archive. These products, derived from single-channel (Landsats 4 -7) and split window (Landsats 8 and 9) techniques, have been coarsely validated by utilizing large water bodies of known temperatures and by utilizing NOAA’s Surface Radiation Budget network (SURFRAD) broadband radiometers over select land targets. However, these datasets supply limited validation data over a limited range of surface emissivities, temperatures, and atmospheric conditions.
In a joint project between RIT and the Seneca Park Zoo, several remote sensing modalities were exploited to assess the biodiversity of the Madagascar rainforest. With the ultimate goal of creating an immersive digital environment that would allow Zoo patrons to experience a Madagascar rainforest, data collections for the physical and sound structure of the rainforest as well as 3D images of some of the many small animals of the rainforest were planned for Ranomafana National Park at the Centre ValBio.
An in-house approach to a water-wave spectrum based surface synthesis and in-water radiative transfer had been demonstrated previously using the DIRSIG 4 framework. New work under this task adapted new user interfaces to manipulate the air-water interface using external tools. This included both geometric considerations (i.e. a geometric description of the height field and localized slopes) and material descriptions (unresolved components of the wind-roughened surface as well as localized material differentiation, such as surface foam).
Lidar remote sensing has shown high accuracy/precision for quantification of forest biophysical parameters needed for ecological management. Although the significant effect of Bidirectional Scattering Distribution Functions (BSDF) on remote sensing of vegetation is well known, current forest metrics derived from lidar data seldom take leaf BSDF into account. Despite the importance of BSDF effects, leaf directional scattering measurements are almost nonexistent, particularly for transmission. Previous studies have been limited in spectrum, lacked models to capture all angles beyond measurements, and did not adequately incorporate transmission scattering. Furthermore, many current remote sensing simulations, which are vital to our understanding of lidar data, assume leaves with Lambertian reflectance, opaque leaves, or apply purely specular or Lambertian transmission. The accuracy of these assumptions and the effect on simulation results are currently unknown. This study captured deciduous broadleaf BSDFs from the visible through shortwave infrared spectral regions, accurately modeled the BSDF for extension to any illumination angle, viewing zenith or azimuthal angle.
This project supports the continued development of DIRSIG, through the ground-up rewrite of version 5, to meet the needs of the user community. Prior work focused on the development of a core radiometry capability with processing efficiency to meet modern requirements in an open, flexible framework. Project development exploited the new interface to enable runtime control of scene objects such as secondary sources (street lights, headlights, etc..), scene dynamics, voxelization and truth collection. Additional work supported the maintenance and documentation of capabilities.
The RIT DIRS Laboratory develops and maintains a high fidelity, radiometrically accurate simulation tool named DIRSIG. This tool generates radiance scenes that mainly represent structures and areas on the ground as seen by a remote sensing instrument. The primary goal of this effort is to optimize DIRSIG capabilities in the maritime regime where imaging the dynamic air-sea surface and transparent water bulk pose different challenges than imaging land. In this effort, DIRSIG developers will work with Arete Associates to extend the model to the maritime regime.
Surveillance of vehicles from remote sensing platforms is a challenging problem when viewing conditions change and obscurations intervene. This project examines the use of multi-modal sensing and dynamic adjustment of tracking and background models to improve tracking performance. The primary mode of sensing for vehicle tracking methods is imaging spectrometry, however other modalities will be considered. Background modeling efforts are exploiting the capabilities of DIRSIG5 to dynamically update scene viewing angle and scene clutter.
The goal of this work is to examine emissivity signature variability (including particle size effects from the signature at the ground level (e.g., powers) into a detection context for study. We seek to answer how such variability impacts the measurement / detection problem. A model currently used at RIT, the Forecasting and Analysis of Spectroradiometric System Performance (FASSP) model, can handle most of the statistical model variations in the LWIR. However, this model only examines variations in radiance in the LWIR.
This project focuses on the development of more reliable estimates of carbon storage in coastal wetland systems. The goal is to use very high-resolution hyperspectral remote sensing imagery and detailed surface models from LiDAR and stereo RGB imagery in combination with highly detailed contemporaneous biophysical ground truth data to construct more accurate assessments of carbon storage in salt marsh and to scale these results to imaging platforms with lower spatial resolution such as satellites and fixed-wing aircraft. Using the mast-mounted hyperspectral imaging system developed by Dr.
This project is in support of a Phase II SBIR effort by Intelligent Automation, Inc., to develop a simulation capability for generating visible and infrared (EO/IR) video imagery of human activity for training algorithms to detect such activity in video sequences. Activity during the 2018-19 academic year focused on the planning and execution of Human View Validation Experiment (HUMVVEX) conducted in Dayton, OH in the fall of 2018. Longwave infrared (LWIR) and midwave infrared (MWIR) video sequences were acquired for several test scenarios including static and dynamic (humans walking) tests. RIT contributed to the effort by calibrating the real imagery and developing a DIRSIG scene of the background environment for use in validating the simulations.
The Sustainable Land Imaging (SLI) program is committed to extend the nearly fifty-year data record of spaceborne measurements of the Earth's surface collected from Landsat' s reflective and thermal instruments. Through the development of a system of spaceborne sensors, and perhaps the inclusion of alternative data sources, the SLI program is interested in identifying cost-effective solutions to acquiring consistent and continuous data to support science applications related to the monitoring of Earth's natural resources.
This project is investigating approaches to application performance driven design of spectral imaging systems. The research combines the use of analytical modeling tools (FASSP) with image simulation (DIRSIG) to identify critical points in the system parameter tradespace with a goal of optimizing the design and operation of tunable spectral imaging systems designed for cubesats. The work has also included development of a novel band selection technique using only the reflectance spectra of the desired target. The figure captures several aspects of the project including a range of materials, some of the simulated images used, and predictions of detection utility.
The PI supports the radiometric calibration of the Landsat thermal band instruments for NASA. Specifically, this involves the on-orbit characterization and calibration of the Landsat 8/ Thermal Infrared Sensor (TIRS) instrument and the pre-flight calibration of the new TIRS-2 instrument for the upcoming Landsat 9 mission. The PI serves as the Deputy Calibration Lead for the TIRS-2 project and is a member of the Calibration and Validation team for the Landsat program.
The two objectives of the research – yield modelling and harvest scheduling- have undergone extensive research over the past year. An RIT-based greenhouse study during winter and spring of 2019 resulted in insightful results regarding both yield and harvest of snap-bean, as a proxy crop. PhD student Amir Hassanzadeh collected high-temporal frequency spectroradiometer data (400-2500 nm) for 48 snap bean plants over the course of a full growing season; this took place using a controlled-environment set-up, thus enabling collection of spectra under a best-case scenario (figure to the right). Our results for the harvest maturity objective show that one could classify based on spectra, as well as spectral and physical attributes between ready-to-harvest and not-ready-to-harvest plants with 76% and 80% accuracy, respectively. For the yield modelling objective, our results from partial least square regression depicts that the best time for yield prediction is 20 days prior to harvest. We aim to investigate harvest modelling and yield scheduling of the snap bean crop of six different cultivars located at a research farm, in Geneva, N.Y. (maintained by our collaborators in Cornell University) during the summer of 2019.
The goal of this project is to capitalize on the potential of using remote sensing technologies identified in through field research in 2018 to support the expansion of conventional and organic table beet production by Love Beets USA, and improve profitability and sustainability.
As the United States Geological Survey (USGS) moves toward delivering higher-level Landsat products to users in their Collection 2 release, the next several years are critical to ensure that these products are developed with highly-calibrated sensor data and that they are adequately validated. The Rochester Institute of Technology (RIT) has a long history of supporting Landsat thermal calibration and in recent years has supported the development & validation of potential surface temperature products for the Landsat thermal archive.
The L'Ralph instrument provides spectral image data from visible to mid-infrared wavelengths for NASA's Lucy mission to the Jupiter Trojan asteroids. The overall objective of this work is to assist in the characterization and the radiometric calibration of the L'Ralph instrument to verify mission requirements and to help develop an image processing chain for the science data. The main efforts will focus on pre-flight instrument-level characterization of the instrument at NASA Goddard Space Flight Center.
Sponsor is looking for methods and techniques to quantify total amounts of NOx in relation to the total amount of explosives that had been consumed. The RIT DIRS Lab will use their drone technology to research this problem.
In its second offering, this year-long course has exposed the students to the real-world considerations that must be under- taken in the design of a new, or use of an existing, imaging systems to be utilized as part of a small unmanned aircraft system (sUAS). This year the students took an in-depth look at the MicaSense RedEdge sensor used on-board the SIRA/ DIRS MX-1 platform. The students performed an end-to-end radiometric characterization/calibration and error analysis as well as a geometric characterization of the 5 individual sensors that compose this multi-camera system.
The African Institute for Mathematical Sciences (AIMS) is an independent graduate education institute founded by Neil Turok. There are several of the institutes spread across sub-Saharan Africa, including Rwanda. AIMS offers Masters degrees in a number of areas of mathematical sciences including topics as wide ranging as climate science, pure mathematics, and big data analytics. Instructors are recruited from international universities and industry to teach their expertise in 3-week modules with a focus on experiential learning.
Professor Vodacek is the GRSS ad hoc liaison for sub-Saharan Africa. To support the development of a greater GRSS presence on the African continent the society has been supporting training activities to foster interest in new members. Professor Vodacek worked with GRSS and the African Center of Excellence for Internet of Things at the University of Rwanda to create and deliver a drone remote sensing workshop.