The invention provides a complex geometry image representation method for a point cloud curve surface. Surface attribute information and geometry information of the point cloud curve surface are mapped into a two-dimension image through a spherical polar coordinate. According to the method, firstly, the Cartesian coordinates [x, y, z] are converted into the spherical polar coordinates [gamma, phi, theta] through the spherical coordinate. Two quantities in the coordinates [gamma, phi, theta] are respectively converted into pixel coordinates, and the other one coordinate is converted into a gray value of the pixel, so that three different geometry images gamma (phi, theta), theta (gamma, phi) and phi (gamma, theta) can be obtained. The geometry information loss can be caused when any one image in the three images is used for representing the point cloud curve surface, so that the complex geometry image representation method for the point cloud curve surface provided by the invention integrally uses the three image conversion modes, and the efficient and precise representation on the point cloud curve surface can be realized. When the method is used, the compression, the editing, the simplification, the model retrieval, the progressive transmission and the like on the point cloud curve surface can be easily realized.