The invention discloses a method for extracting a quadratic surface from a three-dimensional
point cloud. The method comprises the following steps: firstly, fitting a curved surface corresponding to each point by utilizing a K neighborhood
point set of each point, and calculating a curved surface parameter standard deviation corresponding to each point; arranging the point sets according to the sequence of the curved surface parameter standard deviations from small to large; sequentially taking out each point from the sequenced
point set, and when the curved surface parameter standard corresponding to the taken-out point is smaller than a set threshold value and is not marked as a growing point, regarding the point as an
initial seed point; carrying out region growth from the initial seedpoint, marking the
growth point belonging to the curved surface by using the serial number of the current region, and carrying out dynamic fitting on the characteristic parameters of the curved surface until the growth of one curved surface is finished; and repeating the steps to take out the points from the
point set and carry out dynamic fitting until the curved surface parameter standard deviation is greater than the threshold value, and finally extracting the corresponding curved surface according to the mark serial number in the
point cloud. According to the method, the quadratic surfacecan be quickly and accurately extracted from the three-dimensional
point cloud, and the optimal surface parameters are obtained.