Automatic point cloud building envelope segmentation (auto-CuBES) using machine learning

A system for producing a wireframe model of a building envelope from a point cloud representing the building envelope, the point cloud being a result of scanning the building envelope with a light detection and ranging (LiDAR) system, the system comprising:

• a data processing apparatus configured to:
• obtain a point cloud representation of a building envelope, the point cloud including a plurality of points wherein each of the plurality of points collectively represents coordinates, light intensity, and pixel color;
• extract a subset of the plurality of points representing each of the individual facades of the building envelope;
• receive a k-means clustering parameter from a user interface;
• for each façade, using a k-means clustering algorithm, segment the façade into a number of facade sections based on the coordinates and the light intensity of the subset of the plurality of points representing the façade of the building envelope as well as the k-means clustering parameter;
• divide each facade section into three sets of features representing interior, wall, and exterior features, wherein the interior, wall, and exterior features each being associated with certain points of the facade section;
• remove outliers from each facade section;
• produce a building envelope wire frame model from the segmented facade sections having outliers removed, wherein the wire frame model accurately represents dimensions of the building envelope;
• and
• a user interface communicatively coupled with the data processing apparatus and configured to present wire frame model produced by the data processing apparatus.