Advancements in Area-Based Imagery Processing

Transcription

1 Advancements in Area-Based Imagery Processing An integrated Geocoder delivers a streamlined workflow and new capabilities Alexander MacDonald, CARIS, Canada Corey Collins, CARIS, Canada For many years the focus has been on collecting and processing the first bottom return of multibeam sonars. However, the acoustic backscatter acquired by multibeam sonars, as well as side scan sonars, also carries important information about the seafloor and its physical properties. This information provides valuable data to aid in seafloor classification, and important auxiliary information for a bathymetric survey. One necessary step towards this characterization is the creation of more consistent and reliable mosaics of acoustic backscatter. For that, it is necessary to carry out certain corrections on the backscatter to create this final backscatter map. The integration of Geocoder 1 into CARIS HIPS and SIPS 7.0 provides the necessary corrections and analysis tools to efficiently and effectively create and classify high-quality, accurate mosaics. The CARIS strategy has been to deeply embed Geocoder into the HIPS and SIPS code base, which brings numerous advantages. For instance, the core technology for the mosaics is based on existing HIPS BASE Surfaces, giving users a fast, reliable and familiar format for editing and storing their work. This strategy enables processing with Geocoder to be made: faster by software improvements such as multithreaded processes; more efficient through a new, seamless workflow; more flexible through the conjunction of CARIS and Geocoder technologies; and facile due to the streamlined user interface. New Approach A fundamental shift has taken place in SIPS, moving from a line-based to an area-based processing approach. Key to this shift is the introduction of a new tool, Mosaic Editor. It incorporates the backscatter processing and sediment analysis capabilities of an integrated Geocoder engine with the best of CARIS technology to provide users with a seamless bathymetric and imagery processing experience in HIPS and SIPS. Great care has gone into the design of the Mosaic Editor user interface. It boasts context-sensitive controls, reliable

2 default settings, and choice of standard or advanced displays. This is to accommodate a wide range of user needs and allow for a fast learning curve with the software. The workflow is straightforward. After processing bathymetric data, users will open Mosaic Editor to process their imagery, be it side scan, beam-averaged intensities, or time series per beam data (Figure 1). Due to the sophistication of Mosaic Editor and the accuracy of correction by the embedded Geocoder engine, processing any of these imagery sources is surprisingly simple and produces remarkably uniform results. After this, users may create and edit mosaics or use the sediment analysis tool to analyse the sediment angular response. Figure 1: The workflow of Mosaic Editor enables users easily to process side scan, beam-averaged intensities or time series per beam data with the embedded Geocoder engine, providing uniform results for mosaic creation and sediment analysis. Advancements in Imagery Processing A major change for SIPS users is that the raw imagery is processed directly into Georeferenced Backscatter Rasters (GeoBaRs). GeoBaRs allow for initial quality assessment in georeferenced space, and interactive mosaic creation and editing. It is possible to have many GeoBaRs per survey line; a useful feature when dealing with file formats that can have multiple sources of imagery data. For instance, a Reson XTF file may have beam averaged intensity, side scan, and snippet data. Mosaic Editor allows the user to process all of these and even visually compare them if desired. Other possibilities include processing the data at different resolutions or with different options. This new workflow and functionality is a significant advancement in current backscatter and image processing applications. In general, with existing workflows and technologies, users must

3 process each survey line applying corrections in line line by line editors, and hoping that when a mosaic is created that all the data will mesh seamlessly. This is analogous to trying to piece together a puzzle while blindfolded. With Mosaic Editor, and the introduction of GeoBaRs, the user can still process lines individually, but these tools enable the user to visualize all pertinent data as a whole and, if need be, apply further corrections on the fly to ensure that the created mosaic is seamless. Mosaic Editor allows the user to create GeoBaRs from one or more selected survey lines. Within Mosaic Editor, the user has the ability to process imagery data using the Geocoder engine while maintaining the ability to utilise the traditional SIPS engine. The user can control the source data type as well as the corrections to be applied (Figure 2). The correction options include: Auto Gain correction (transmit power, receiver gain, and pulse width are normalized to correct banding artefacts); Automatic Time Varying Gain correction, Beam Pattern correction (removes the angular artefacts innate to the transducer); Angle-Varying Gain correction (removes other angular artefacts, especially the angular varying response of sediment); and finally Despeckle (which reduces noise in the imagery). Figure 2: The Mosaic Editor allows the user to control the processing of GeoBARs as well as provide controls for Geocoder s sediment analysis tools and on the fly editing tools for GeoBARs or Mosaics.

4 The Geocoder engine has other corrections that are applied automatically as required. For instance, when a BASE Surface is designated as the source DTM, Grazing Angle correction is automatically performed during imagery processing Efficient Quality Control The array of automated geometric and radiometric corrections in Geocoder is the perfect complement to the advanced manual correction tools in SIPS. By initially processing the survey imagery using the Geocoder engine, the user may quickly produce a detailed mosaic (Figure 3). Figure 3: Fully integrated Geocoder technology produces high quality image mosaics (Data from Shallow Water Conference 2005). If there are areas of concern arising from inspection of the mosaic, the user may elect to examine the data more closely. The georeferenced, corrected data may provide the necessary clues to reprocess some or all of the data. If a deeper analysis is required, a time-based, waterfall style editor is integrated into Mosaic Editor. The Waterfall Editor is capable of ping based editing, processing and analysis that will be immediately familiar to SIPS users. It features a signal display window, contact editing tools, numerous correction tools and an area based zoom window. Once the imagery has been processed and verified, the user may perform Live Editing of GeoBaRs and Mosaics using the context-sensitive controls on the Edit tab. The user may also access the abilities of Geocoder s Angular Range Analysis tool through the Sediment Analysis interface in Mosaic Editor.

5 Creating and Editing Mosaics The introduction of GeoBaRs provides significant new opportunities for creating mosaics in SIPS. A GeoBaR is a common container for georeferenced, processed imagery data in SIPS. Due to this conformity, mosaics can be constructed using processed data from any number of related surveys, independent of equipment type or logging format. Additionally, mosaics may be created from GeoBaRs of different resolutions, placing no limitations on their usage. How does one now create a mosaic in SIPS? Mosaic creation begins with the Create Mosaic Panel. The Create Mosaic panel is simple and provides the User with controls to create mosaics based on five industry proven methods, as well as the ability to select a desired colour map for the mosaic (Figure 4). Figure 4: Create Mosaic Panel As with existing workflows in SIPS, a Field Sheet is required to create the mosaic. The Create Mosaic panel provides a starting point for the user. Once the mosaic is created; the ability to make modifications is available in the Edit tab of the Mosaic Editor with regards to individual components (GeoBaRs) of the mosaic, or the mosaic as a whole. The multi-step mosaic wizard from previous versions has been replaced with this Edit panel that intuitively knows what you are trying to do and presents the appropriate controls to do it (Figure 5).

6 Figure 5: The Edit panel to the left above contains controls to edit GeoBaRs while the Edit panel to the right contains controls to edit the mosaic as a whole. Editing options include contrast / brightness controls, Z-ordering, blending, and add / remove / modify options for mosaic components. All of these controls work on-the-fly for live editing of mosaics in post-processing (Figure 6). Figure 6: Before and after images produced from the live edit of a mosaic using on-the-fly post-processing options (Data from Shallow Water Conference 2005).

7 Since a mosaic might not be completed in one session, the edit controls will remember everything about the specific settings used each time the mosaic is selected for editing. Furthermore, if a mosaic is modified to the point where the results are now undesirable, due to the maintaining of the original attributes of the mosaic, the mosaic can be reset to its original state. Sediment Analysis One of the most innovative aspects of Geocoder is its sediment analysis capabilities. This is made possible by the accurate removal of acquisition artifacts found in the source data and by the sediment angular response models built into Geocoder. These models were implemented in Geocoder by Dr. Fonseca and are based on years of published research in this area. This is a crucial point because it is these models that drive the analysis, rather than any sort of curve fitting algorithm. This powerful tool, known in Geocoder as Angular Range Analysis (ARA), is realized in SIPS as the Sediment Analysis Tool (SAT). The SAT was designed to provide the complete capabilities of the Geocoder ARA engine in an easy-to-use interface that accommodates the varying needs of SIPS users. The Sediment Analysis Tool (SAT) is used primarily to analyze corrected backscatter returns. In its current implementation, GeoBaRs are divided into user-defined, regularly-spaced patches for port and starboard returns. For any given patch, the average signal response is shown on the SAT graph window (Figure 7). Figure 7: Sediment Analysis Window displays the average signal response for any given patch.

8 When analysing, the signal traces are joined by the Total Backscatter Model trace as the Geocoder ARA engine attempts to classify the response. The determined average grain size is cross-referenced to a lookup table in order to provide the textural response (e.g. silty clay). This classification table is configurable in the SIPS implementation. The analysis also provides a confidence value as well. In the advanced mode of the SAT, the user has the freedom to interact with the model, adjusting the parameters and seeing the graph s backscatter traces update automatically. The user may ask Geocoder to analyze the angular response they have entered to test the corresponding textual representation from Geocoder. This mode is useful for experts who may wish to specifically configure the Geocoder responses, or who may wish to manually investigate certain angular response properties. Advantages of Embedding Geocoder in SIPS This unique approach of using Geocoder as an embedded engine, rather than a standalone application (Figure 8), has proven to have significant merits. Figure 8: Data such as navigation and processed depths are routed to the Geocoder engine during processing rather than from the source for greater production efficiency. First among them is that it has proven to be significantly faster during benchmark testing. Necessary data, such as navigation and processed depths, are automatically routed to the embedded Geocoder engine when processing imagery rather than being read from the source

9 data file. Also, the Digital Terrain Models (DTMs) needed by Geocoder require only that the user designate a BASE Surface for imagery processing. Second, GeoBaRs are built on BASE Surface technology and thus provide benefits such as data caching, multiple data layers, and most importantly, scalability. With the introduction of the new CARIS CSAR framework to GeoBaRs later this year, fully editable, gigapixel mosaics will be realizable. Planned integration with CARIS Bathy DataBASE will enable the proper management of Mosaics and GeoBaRs as well as promote efficient data sharing between groups. Third, a notable advantage in some environments has been the ability to create, share, recall, and apply process templates for performing work in SIPS. This allows for routine operations to be automated, and also permits organizations to standardize operations and workflows. The SIPS Template Wizard (STW) has been augmented to support Geocoder processing in SIPS. Conclusions The strengths of Geocoder are certainly in its array of detailed backscatter corrections and its accurately modeled seafloor characterization algorithms. The challenge then is to blend these capabilities into a coherent workflow, design a clear yet powerful user interface, and provide the scalability, stability and efficiency needed by organizations of any size. CARIS HIPS and SIPS 7.0 has succeeded in meeting this challenge and a new benchmark in backscatter processing has been set.

10 Acknowledgements 1. Geocoder was developed by Dr. Luciano Fonseca and is licensed to CARIS by the University of New Hampshire. Further Reading Fonseca, L., and Mayer, L.A., 2007, Remote estimation of surficial seafloor properties through the application Angular Range Analysis to multibeam sonar data. Marine Geophysical Researches, p (Refereed Publication) Fonseca, L., and Calder, B., 2005, Geocoder: an efficient backscatter map constructor. Proceedings of the U.S. Hydrographic Conference 2005, San Diego, C.A. MacDonald, A. J., and Collins, C., 2008, Taking Geocoder to Work: Making Geocoder part of your Daily Routine: Proceedings of the Shallow Survey Conference 2008, Portsmouth, N.H. Masry, M., and Collins, C., 2008, Scaling Bathymetry: Data handling for large volumes. Proceedings of the Shallow Survey Conference 2008, Portsmouth, N.H. Authors Alexander MacDonald, B.Sc.Eng Development Manager SIPS CARIS Fredericton, New Brunswick, Canada Phone: +1 (506) Fax: +1 (506) URL: Corey Collins HIPS/SIPS/Notebook Product Manager CARIS Fredericton, New Brunswick, Canada Phone: +1 (506) Fax: +1 (506) URL: