The invention provides a Chebyshev feature extraction method of a 3D model based on rays. WPCA preprocessing is carried out to rectify the 3D grid model; a Chebyshev spherical surface of the model is established, the surface is divided into longitude and latitude areas, the 3D grid model rectified by WPCA preprocessing is arranged in a unit sphere of the Chebyshev spherical surface, and a sampling window is designed; triangular densification is carried out on patches in the sampling window, dense points are processed to maximal and minimal sampling distances in the area, a 2D feature signal, namely, an original Chebyshev feature signal, is obtained; and sparsification is carried out on the original Chebyshev feature signal, 2D compressed sensing processing is carried out, and a low-dimensional compressed sensing feature FCSCR is obtained. The robustness of the extraction method is enhanced, the 3D model is described in a more elaborate manner, and the high-efficiency dimension-reduced compressed sensing feature can be obtained.