Hierarchical Image Segmentation (HSEG)

Image data is subdivided into overlapping sections. Each image subsection has an extra pixels forming an overlapping seam in the x, y, and z axes. The region labeling these overlapping seams are examined and a histogram is created of the correspondence between region labels across image subsections. A region from one subsection is then merged together with a region from another subsection when specific criterion is met. This innovation’s use of slightly overlapping processing windows eliminates processing window artifacts. This innovative approach can be used by any image segmentation or clustering approach that first operates on subsections of data and later combines the results from the subsections together to produce a result for the combined image.

NASAs Goddard Space Flight Centers method for the extraction of Lunar data and/or planetary features is a method developed to extract Lunar features based on the combination of several image processing techniques. The technology was developed to register images from multiple sensors and extract features from images in low-contrast and uneven illumination conditions. The image processing and registration techniques can include, but is not limited to, a watershed segmentation, marked point processes, graph cut algorithms, wavelet transforms, multiple birth and death algorithms and/or the generalized Hough Transform.

The High-Resolution Commercial Validation Products API generates VHR products to support scientists and missions that include two primary foci. The first focus is on-demand VHR regional mosaics. Systematic ortho-rectified and co-registered multi-temporal, panchromatic and unsharpened multi-spectral imagery compiled as user defined regional mosaics will allow for spatially continuous and temporally consistent reference. Such reference will provide an easily accessible calibration and evaluation dataset for scientists. Surface reflectance data from VHR imagery is derived using the Multi-Angle Implementation of Atmospheric Correction (MAIAC) algorithm and compared over pseudo-invariant calibration sites for cross-calibration to minimize the effects of topography, view angle, date and time of day of collection. The API includes a process for mosaicking and normalizing ortho-rectified images to create scientific data products useful for many programmatic activities, including biodiversity, tree canopy closure, surface water fraction, and cropped area for smallholder agriculture. The second focus is On-Demand VHR digital elevation models (DEMs). Systematic processing of available along- and cross-track stereo VHR imagery is used to produce VHR DEMs. A systematic DEM co-registration approach has been applied to generate products with horizontal and vertical accuracy that can support missions and a number of different science programs. These include studies on the cryosphere, hydrology, and the biosphere.

NASA's GONASA technology is a mathematical formula / algorithm built around creating a grid composed of equal-area cells that span the entire visible hemisphere of a spherical object. Traditional longitude and latitude grids produce cells that diminish in size toward the poles due to convergence of longitudinal lines. GONASA circumvents this problem by carefully adjusting the latitude increments, resulting in a network of truly equal-area cells. This adjustment ensures that any feature observed on the spherical surface is accurately represented, regardless of its location. To implement GONASA, the spherical surface is first segmented into discrete latitude bands or rings, each chosen to encompass an identical surface area. Within each ring, longitude divisions maintain equal cell areas, creating a uniform Cartesian grid. The result is a consistent, distortion-corrected matrix suitable for automatic computation, enabling simplified, efficient, and accurate measurements of spatial characteristics such as feature area, centroid location, perimeter, compactness, orientation, and aspect ratio. GONASA grids are computationally efficient and readily adaptable to a range of data processing workflows, from spreadsheets to sophisticated data analysis frameworks like Pandas data frames in Python. Due to their consistent cell sizing and straightforward indexing, GONASA grids facilitate automation, enabling rapid, high-volume data processing and analysis, essential for modern remote sensing and planetary missions that require immediate, reliable data analysis in limited-bandwidth communications environments. At NASA, GONASA has already been successfully implemented to study images of Titan (e.g., mapping its clouds) taken by the Cassini space probe.

The innovation improves upon the performance of passive automatic enhancement of digital images. Specifically, the image enhancement process is improved in terms of resulting contrast, lightness, and sharpness over the prior art of automatic processing methods. The innovation brings the technique of active measurement and control to bear upon the basic problem of enhancing the digital image by defining absolute measures of visual contrast, lightness, and sharpness. This is accomplished by automatically applying the type and degree of enhancement needed based on automated image analysis. The foundation of the processing scheme is the flow of digital images through a feedback loop whose stages include visual measurement computation and servo-controlled enhancement effect. The cycle is repeated until the servo achieves acceptable scores for the visual measures or reaches a decision that it has enhanced as much as is possible or advantageous. The servo-control will bypass images that it determines need no enhancement. The system determines experimentally how much absolute degrees of sharpening can be applied before encountering detrimental sharpening artifacts. The latter decisions are stop decisions that are controlled by further contrast or light enhancement, producing unacceptable levels of saturation, signal clipping, and sharpness. The invention was developed to provide completely new capabilities for exceeding pilot visual performance by clarifying turbid, low-light level, and extremely hazy images automatically for pilot view on heads-up or heads-down display during critical flight maneuvers.