Application of Landscape Mosaic Technology to Complement Coral Reef Resource Mapping and Monitoring

Dr. Pamela Reid | University of Miami

RC-1333

Objective

Recent declines in coral reefs across the globe underscore the need for new scientific tools to better understand ecological patterns and rates of change. Of immediate interest to the Department of Defense (DoD) are measurements that detect changes within reef systems. For coral reefs, successful and legally defensible monitoring of reef condition requires estimates of basic ecological parameters, such as live coral cover, species diversity, and mortality/recruitment rates. Presently, such parameters are typically measured during field surveys using trained divers. Airborne or satellite-based remotely sensed data are not currently able to reliably quantify coral condition at the required level of detail. The most appropriate strategy for reef monitoring is detailed analysis of meter-scale plots at high spatial resolution. The development of underwater landscape mosaicing capability addresses an emerging need to augment diver surveys in an effort to efficiently inventory and monitor large areas of DoD-held coral reef resources.

The primary objective of this project was to develop technology that will increase the speed and repeatability with which reef plots can be mapped and inventoried. Specifically, underwater video was used to create two-dimensional (2-D) landscape mosaics of reef plots in a highly automated way. The goal was to construct spatially accurate landscape video mosaics of reef plots and to extract meaningful ecological indices of reef condition from these mosaics. As secondary objectives, this project explored two techniques to assist or automate classification of underwater imagery: (1) three-dimensional (3-D) reconstruction of specific reef features and (2) high resolution multispectral imaging. 

As an extension to the original proposal, the primary objective was broadened to include expanded mosaicing capabilities and streamlined processing. Specifically the scope was expanded to reduce the impact of sun flickering interference on mosaic processing, increase benthic resolution, and improve image blending to create seamless mosaics.

Back to Top

Technical Approach

Spatially accurate 2-D landscape mosaics, about 20 m by 20 m in extent with millimeter-scale resolution, were constructed using a mosaic algorithm.  This algorithm involves three major steps: (1) estimating image motion by matching sequential video frames, (2) using global alignment to match non-sequential frames, and (3) blending to render the seamless mosaic image.

To make 3-D measurements, a 3-D map of an object or region of interest was constructed. Feature-based 3-D reconstruction of single objects was done using an underwater stereo camera to acquire overhead stereo images of the survey area.  Then the stereo image pairs were processed to form dense depth maps.  Points were selected near the top and bottom of each object of interest to determine the heights of the objects.

The 2-D method was further extended to deal with a two-camera configuration, comprising a video and a high-resolution still camera. The extended method performs the registration of stills against the video frames to find the adequate placement of the stills in the mosaic.

Back to Top

Results

An important component of this project was to assess whether landscape mosaics retained the strengths of diver-based coral reef assessment. Five classic indicators of reef condition were considered in addressing this question: (1) percent benthic cover, (2) diversity index, (3) juvenile coral density, (4) disease/bleaching/partial mortality, and (5) coral colony size.

Benthic indicators extracted from the mosaics showed that performance was considered excellent (i.e., on a par with divers). Sizes of coral colonies measured from both first and second-generation mosaics did not significantly differ from diver-based measurements, and thus mosaic measurements were assessed as excellent. As a result, species-level identification was possible.

The resulting mosaics excel in several areas that are otherwise limited by poor performance using traditional, state of-the-art, reef monitoring techniques:

  • Mosaics provide a landscape view of coral reefs that has previously been unobtainable.
  • Mosaics are superior tools to track patterns of change over time.
  • Mosaics have a spatial accuracy on the order of centimeters to millimeters.

As a consequence of the capabilities above, mosaics are superior tools for monitoring disease, bleaching, and partial mortality—all important indicators of reef health.

Because of habitat complexity, monitoring tools that document the 3-D topography can provide valuable information on reef structure and function. Such information is not commonly obtained by standard monitoring methodologies for measuring coral cover as the proportion of the bottom occupied by corals in a planar view. More precisely, utilizing this tool for the 3-D size measurements, the user can more readily acquire the minimum technical knowledge for assessing the performance of the underlying methods constituting the entire system. This investigation has demonstrated that the 3D system is a useful tool to determine object sizes automatically from stereo views.

This technology transitioned to an Environmental Security Technology Certification Program (ESTCP) demonstration project: High Resolution Landscape (2-D) Mosaics for Improved Coral Reef Monitoring Capability ( RC-201021).

Back to Top

Benefits

The resulting landscape mosaics enable users to extract increased reef health information in complex reef habitats and provide a better overall product with increased value to coral reef monitoring users for change detection analyses. These are essential capabilities for the legally mandated environmental documentation necessary for conducting military operations and could provide decision-makers with crucial information necessary to maintain compliance with relevant statutes, regulations, and executive orders.

Landscape mosaics developed under this project provide scientists and resource managers with novel, powerful survey tools. Project deliverables include software to generate mosaics, tools to facilitate viewing the mosaics and assist in the extraction of ecological information, and user manuals. Landscape mosaics can now be used to routinely map large underwater areas at high spatial resolution (on the order of 400 m2 at sub-mm resolution), resulting in spatially accurate, landscape views of the bottom that were previously unobtainable. These landscape mosaics will be useful for DoD reef monitoring requirements and open doors for new applications in reef mapping and change detection.

Back to Top

2009

Points of Contact

Principal Investigator

Dr. Pamela Reid

University of Miami

Phone: 305-421-4606

Program Manager

Resource Conservation and Resiliency

SERDP and ESTCP

Share