Control and quantification of visible emissions is a challenge for Department of Defense (DoD) facilities. Some emission sources require in-stack monitors as part of a facility's permit; however, because of their installation and maintenance costs, in-stack monitors are a less common method for quantifying visible emissions than Environmental Protection Agency (EPA) Reference Method 9 (Method 9). Method 9 requires that a certified human observer view the plume against a background to quantitatively determine opacity. The use of EPA-certified observers for measuring opacity is expensive because it requires each observer to attend two training courses per year and larger DoD installations can require numerous EPA-certified observers. Beyond the costs, opacity measurements from human observers are subject to inaccuracies and lack reproducibility. Digital methods, in addition to offering accuracy and repeatability, can be operated remotely.

The objective of this SERDP Exploratory Development (SEED) project was to develop a small, accurate, and robust system for determining opacity, based on the analysis of digital images. Enhancements to digital methods that were explored include having these methods work under a wide range of conditions, function for a wide range of commercially available digital cameras, take into account principles of light attenuation, not require the introduction of a physical background behind a plume, and operate with the next generation of digital image-based opacity measurements.

Investigators developed image-processing algorithms to determine opacity from a digital image. The accuracy of the algorithms was validated using an extensive database of digital images from Hill Air Force Base, Utah, for which corresponding Method 9 and transmissometer results were available. Specifically, investigators evaluated the new algorithms for different plume colors, intensities, backgrounds, and environmental conditions. The work relied on a contrast-based strategy to quantify light attenuation. For each image, a user selects a region of interest containing both the plume and background. The image-processing algorithms segment the image into plume and background regions and use differences in local contrast to determine opacity. A challenge for opacity estimations lies in the variations in the contrast of background; however, investigators incorporated techniques such as averaging to overcome these variations.

A digital system for determining opacity offers significant benefits to DoD in its potential to reduce environmental compliance costs while providing accurate, reproducible measurements. In addition, this small, portable, and flexible system will provide a permanent record of its results. The system will enable DoD to maintain and, in some cases, expand mission-critical operations with a reduced risk of regulatory restrictions. Beyond mitigating any potential reduction in facility activities, implementation of the digital opacity methods can reduce DoD's reliance on Method 9-certified human observers and costly in-stack monitoring applications. (SEED Project Completed - 2007)