remote sensing image-processing models
Endpoint Environmental LLC specializes in building image-processing models within commercial-off-the-shelf software packages. Our goal when designing image-processing models: increase interpretation efficiency by reducing the number of steps conventionally required to generate geospatial data from remotely sensed imagery. This efficiency saves our clients long-term field inspection time and monetary resources. The accuracy of our models typically ranges within the 80th and 90th percentile. We have built models to map urban-sprawl and land-use conversion, aquatic invasive plant, and waste tire-piles.
Feature Extraction: aquatic invasive plants
Egeria densa, a submergent aquatic invasive plant species, has been growing in the Sacramento-San Joaquin River Delta for decades. We built a model to map Egeria densa in the Delta quickly and efficiently. This level of map production is useful for decision-makers who require an understanding of an invasive plant’s spatial distribution or change in coverage annually or over significant periods of time. Accurate knowledge of the spatial distribution of an invasive species can be used to target chemical and physical eradication treatments and possibly reduce the cost of managing invasive plant species paid for by the U.S. taxpayer.
Change Detection: urban sprawl and land-use conversion
Where does an urban development go when it sprawls? How fast does urban sprawl occur? What land-use/land-cover types are most often destroyed when developing urban areas? In order to address these questions, we created a semi-automated image-processing model. Remotely sensed imagery is cleaned for noise and classified into land-cover types. The data is then processed through a custom-made model to determine where the sprawl occurred, the rate of change, and which land cover types were converted to urban land-use. The model is designed to process any two or more annually successive images of a given location worldwide.
