Local similarity measure for landslide detection and identification in comparison with the image differencing method
In this article, a new simple method of landslide detection and identification is proposed. It is based on the use of local mutual information and image thresholding. A binary change image is then produced. Connected component analysis is used to identify the connected regions. Landslides are identi...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Published: |
Taylor & Francis
2010
|
| Summary: | In this article, a new simple method of landslide detection and identification is proposed. It is based on the use of local mutual information and image thresholding. A binary change image is then produced. Connected component analysis is used to identify the connected regions. Landslides are identified as the largest connected regions in this image. Mathematical morphology is used to approximate the landslide region. This method is simple and suitable for the detection of large changed regions where the ratio of the unchanged to changed pixels in the image is approximately one to a few tens. Compared to the image differencing method, this method gives more reliable results. |
|---|