LAUSR

In vector model application fields (such as high-capacity, high-precision GIS vector mapping, and CAD drawings), security technologies for authentication and retrieval have recently become necessary. In this paper, we propose a unique, multi-scale, curvature-based perceptual vector model hashing method with hierarchical authentication, superior robustness, and superior security. In our hashing method, multi-curvature activity energies on multi-scale models are obtained for all polylines and polygons using radius curvature, turning angle curvature, and Gaussian curvature, which are invariant to rigid motion and robust to shape deformation. Following this, we generate multidimensional binary hash values by random mapping with partial exponential Bell polynomials and by Lloyd–Max quantization. Experimental results confirm that our method reduces the distance error of object attacks by 0.001–0.068 and improves both the unique probability by about 0.014, and the differential entropy by 0.875–2.149 compared with a conventional method.