Ground point cloud extraction method integrating multi-level progressive strategy and unsupervised learning

The invention discloses a ground point cloud extraction method integrating a multi-level progressive strategy and unsupervised learning, and the method comprises the following steps: S1, converting point cloud into a depth image, and obtaining two-dimensional grid data; S2, adopting a median denoising method to remove noise data in the depth image; S3, setting a filtering window scale range, and carrying out size marking on the grid data by adopting morphological high-cap operation; S4, setting area and roughness constraint conditions, detecting the maximum building size, and determining an optimal filtering window at the same time; S5, calculating the gradient change of each local terrain region according to a morphological filtering result, and setting a filtering threshold as a linear function of the gradient change; and S6, filtering is carried out point by point according to the point primitives according to a self-adaptive changing filtering threshold value. According to the method, a multi-level progre