The invention relates to a method for extracting buildings with uniform spectral characteristics from
remote sensing images. The method includes the following steps: firstly, utilizing a likelihood function of total variation of neighborhoods to calculate a value LS of each pixel inputted into the
remote sensing images; secondly, performing binarization
processing; thirdly, performing marking; fourthly, calculating the
center of mass of a marked region; fifthly, performing
region growing segmentation; sixthly, performing preliminary building extraction; seventhly, acquiring a segmentation object sequence S; eighthly, selecting a training sample; ninthly, extracting
texture feature information; tenthly, performing classification. The problem of low accuracy of building extraction in high-spatial-resolution
remote sensing images is solved, full
automation is achieved, and the method is used for
data acquisition and automatic updating of remote sensing image mapping and geographic information systems.