Statistics – Computation
Scientific paper
Dec 2010
adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2010agufmin33b1303e&link_type=abstract
American Geophysical Union, Fall Meeting 2010, abstract #IN33B-1303
Statistics
Computation
[1914] Informatics / Data Mining, [1994] Informatics / Visualization And Portrayal, [5464] Planetary Sciences: Solid Surface Planets / Remote Sensing
Scientific paper
NASA’s two Mars rover spacecraft, Spirit and Opportunity, have collected more than 4 years worth of data from nine imaging instruments producing greater than 200k images. To date, however, the potential ‘field’ data that these images represent has remained relatively untapped because of a lack of software with which to readily analyze the images quantitatively. We have developed prototype software that allows scientists to locate and explore 2D and 3D imagery captured by the NASA's Mars Exploratory Rover (MER) mission robots Spirit and Opportunity. For example, using our software, a person could measure the dimensions of a rock or the strike and dip of a bedding plane. The developed software has three aspects that make it distinct from existing approaches for indexing large sets of imagery: (1) a computationally efficient image search engine capable of locating MER images containing features of interest, rocks in particular, (2) an interface for making measurements (distances and orientations) from stereographic image pairs and (3) remote browsing and storage capabilities that removes the burden of storing and managing these very large image sets. Two methods of search are supported (i) a rock detection algorithm for finding images that contain rock-like structures having a specified size and (ii) a generic query-by-image search which uses exemplar image(s) of a desired object to locate other images within the MER data repository that contain similar structures (i.e. one could search for all images of sand dunes). Query by image capabilities are made possible via a bag-of-features (e.g. Labeznik et. al. 2003; Schmid et. al. 2002) representation of the image data which compresses the image into a small set of features which are robust to changes in illumination and perspective. Searches are then reduced to looking for feature sets which have similar values; a task that is computationally tractable providing quick search results for complex image-based queries. Images recorded by the MER pancam instrument often include an (x,y,z) position information for each image pixel which can be localized to the global Mars Body Fixed (MBF) coordinate system. This allows position, distance, and orientation measurements from objects imaged by the pancam as stereographic image pair. Analysis tools paired with our search capabilities allow researchers to obtain measurements directly from MER imagery, to visualize these measurement(s) and the data in 2D and in 3D, to collate numerous measurements for statistical analysis and to estimate the uncertainty of each recorded measurement. The combination of data mining functionality in conjunction with analysis and visualization tools provides MER researchers unprecedented abilities to extract quantitative information from this important data not provided by current MER software systems. Our group will be employing the software to test recent hypotheses (e.g. McFadden et al., 2005; Eppes et al., 2010) related to the efficacy of diurnal heating and cooling in cracking rocks on the surface of Mars.
Eppes Martha C.
Willis Allan
Zhou Baoguo
No associations
LandOfFree
Collecting field data from Mars Exploration Rover Spirit and Opportunity Images: Development of 3-D Visualization and Data-Mining Software does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with Collecting field data from Mars Exploration Rover Spirit and Opportunity Images: Development of 3-D Visualization and Data-Mining Software, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Collecting field data from Mars Exploration Rover Spirit and Opportunity Images: Development of 3-D Visualization and Data-Mining Software will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-1502282