Keeping informed of the collieries exploitation is of great importance for the benefit of our country. Currently, with the
high spatial resolution remotely sensed imagery, we are able to monitor the open air collieries directly without in situ
inspections. This paper presents the application of active contour models(snakes) for semi-automatic extraction of the
contour from the typical collieries in remote sensing images. After carefully examined the characteristics of the
collieries, we improved the active contour model. The boundary of the collieries is not very clearly on the images, and
the influence of the random image noise is extremely large. Therefore, improving the image power of Snake model is
necessary. As the Snake model can only work on panchromatic images, we first fused the multi-spectral image to make
full use of the spectral information contained with in the image. Then we used Canny edge detector which is anti-noise to
extract the features. At the same time, a gauss filter is performed to the edge image to enlarge the envelope of the edge.
We found that the image power calculated from the processed image is more efficient than it from the original image. A
software package build in ArcGIS has been made based on this method.
|