抄録
This paper describes a slope failure detector based on mathematical morphological operations. Multisensor data, LANDSAT TM, ADEOS AVNIR, SPOT HRV XS and HRV-Panchromatic, are chosen as the original images. The proposed detector consists of morphological fundamental operation such as dilation and erosion. The formula was applied to the detection of earthquake-induced slope failures in Rokko Mountain side triggered by the Hyogoken-Nambu earthquake. The results lead two major conclusions as follows;
(1) The accuracy of proposed slope failure detector depends on the resolution and high-resolution data made by wavelet transform is effective for the input data. Best result of proposed method showed almost perfect for failures of over 1000m2.
(2) The accuracy of detection depends on the band (wavelength) of the input data rather than the resolution.
