A Man-Machine Partnership for Map Production: An Application of Image Classification and Auto-Vectorization in Charting Coastlines

by Kenneth White, Texas A & M Univ, College Station, United States,
Rendel Clark, Texas A & M Univ, College Station, United States,
Andrew Rost, Texas A & M Univ, College Station, United States,



Document Type: Proceeding Paper

Part of: Coastlines of the Gulf of Mexico

Abstract:

The semi-automatic production of coastal area maps through computer processed digital images is described. Two sources of digital imagery were addressed: satellite and aircraft mounted multispectral scanning systems and scan digiting aerial photographs which produces a digital resampling of the original analog source. All digital imagery was exposed to an image classification algorithm for feature recognition. The resulting thermatic image was processed by an auto-vectorization algorithm which captured the coastline as a series of vectors. Photographic evidence clearly supports the computer generated map as being cartographically correct. The software produced in the study is not limited in scope to coastal mapping. Subsequent modifications to the system have enjoyed successes in fields as diverse as identifying wildlife habitats and estimating species population; crop yield by examining the number, size and vigor of fruit trees; and porous rock oil reserves by examining the pores on scan digitized photographs of thin slices of oil bearing rock.

Subject Headings: Computer vision and image processing | Mapping | Coastal processes | Aerial photography | Algorithms | Shores | Rocks

Services: Buy this book/Buy this article

 

Return to search