Abstract- 
There is imprecision in the stage of construction because of manual quality control in the construction procedures. Forming and processing point clouds image of as is truss part with its CAD data are researched to identify and repair the geometric discrepancies and imprecision. Approved suggested technique of processing building model was done.
Instead of manual processing of the 3D models points. BPNN architecture give the best coordinates processing training. It is applied to optimize the points forming process between the input laser scanned coordinates of the as-built model with the output true CAD coordinates. RMSE in x, y, xy, z and space vector R of all processing tasks are computed acceptably.