An automatic change detection and recognition system for forestry
A new method for change detection and identification, AutoChange, was developed and tested. AutoChange uses as an input 2 images, acquired at different times, and a parameter list given by the user. The method does not apply any sophisticated image calibration. Change detection and identification are separated. The output is a 5 channel image estimating the change magnitude and characterizing the changed and unchanged areas. The method carries out the actual change analysis using homogenous units selected from the images. Only in the ultimate phase is the whole image classified. No ground data are used to produce the analysis results. In test using Landsat thematic mapper data, clear cuts could be detected and identified reliably. The method was also able to give information on forest damage because the intensity change was consistent on damaged areas despite its minor magnitude.
Bibliographic Reference: Article: Remote Sensing of Environment (1995)
Record Number: 199512109 / Last updated on: 1996-01-15
Original language: en
Available languages: en