T. Tamm, K. Remm, and H. Proosa (Estonia)
Local statistics, texture, forest remote sensing
The need to include areas around the target location into remote sensing predictions is being increasingly stressed. This paper introduces the LSTATS software for calculating local statistics both in the local kernel and as well as the segmented portions of an image. Ten special features not commonly used by the remote sensing community were found and described, and their potential application in forest remote sensing was presented. The “Weighted Moran's I”, “Homogeneity of neighbours” and “Difference between centre and boundary” are examples of distinctive local statistics. Local statistics considered in this paper can be most helpful in forest remote sensing systems for distinguishing shadowed management passages, spruce canopies, groups of tree crowns and clearings in forests.
Important Links:
Go Back
