A Detection Method of New Buildings Based on High Resolution Remote Images

Traditional change detection methods based on the pixel layer didn´t explore the features such as texture, structure, shape and the accuracy was influenced by the the phenomenon of the same surface feature with different spectrums or different surface features with the same spectrum and salt and pepper noise. This paper adopted the object-oriented change detection method, the late-temporal was segmented by the multiresolution segmentation, the building objects were extracted, then according to the features of spectrum and shape the shadow objects were extracted for compensating the missing buildings. By judging whether the values of spectrum and texture between the building objects in the late-temporal image and the patchs which had the same location and range with the late-temporal buildings in the first-temporal image were different, in that case the late temporal building patchs were new additions.