A flexible selective cryptosystem for meshed 3D textured animated objects and images using cellular automata

Abstract

One of the many issues in three dimensional (3D) objects is their security. One of the core difficulties in securing them is gathering between simplicity and reliability while designing the cryptosystem, especially with their huge information. Studies on securing 3D objects can be divided into four main branches: hologram, voxels, point cloud, and mesh. However, these studies seldom focus on the mesh, although it is the most applied in applications. In addition, algorithms that focus on mesh lacks flexibility in their techniques and in designing secret keys, and they are limited to encrypt nude single rigid object or to encrypt texture separately from its object. Therefore, there is a need to design a flexible and reliable cryptosystem for 3D meshed objects, and to jointly encrypt (texture, texture map, and animation). In this research, two different 3D animated objects were encrypted, the first is animated by kinematics, and the second is animated by rigging. Encrypting both kinds of animated objects essentially based on (FcCA), where FcCA is a new system, called Flexible Cryptosystem based on Cellular Automata (CA), FcCA was proposed due to reasons that in cryptography, the previous research generally appears ambiguous and complex to the novice researcher due to the use of complex mathematical equations and rules of cryptography. Therefore, FcCA proposed five novel techniques: pure random cellular automata (configuration and boundary conditions), g-th order memory independent-cell, two parts of key: (HCA, Ps), and dynamic start locations (cDL). An Improved FcCA (iFcCA) overcomes the drawbacks of FcCA, and fully encrypts 3D textured objects with their texture maps. Where iFcCA presents two new techniques: random dynamic intersected start points (rDL) and selective masking for plain data histogram (sHist). Then, (SCEAT3D) proposed thirty descriptors for selecting and compressing animated objects, it enhances iFcCA by three techniques: identifying the rigid object by one meaningful point (Mp), summarisation by selecting rigid key objects (Megalooikonomou et al.), and compression of selected rigid key objects (CKO). FcCA and iFcCA were compared with the-state-of-the-art image and 3D object encryption algorithms. FcCA and iFcCA have the largest keyspace > 226,818,167 besides high key sensitivity, both have close or superior results under standard criteria comparing with others; where iFcCA has the lowest PSNR, MAE > 81.6%, secondbest score for entropy, standard UACI, strong against noise and occlusion robustness, totally uniform histogram after encrypting. Overall, the properties of both demonstrated their effectiveness as cryptosystems for images and 3D objects. Moreover, implementation and analysis of SCEAT3D showed that it can be a competitor for the human user how are interested in the field, where the highest average for finalrate by a human was (61.2%) which obtained only once time, while three proposed descriptors surpassed this average with (63.3%, 61.6%, 61.5%). Moreover, it reduces encrypted data while keeping the quality and the integrity of the sequence of motion very smooth, and consequently, reduces the time and the space complexity of the whole system. The findings from this research confirmed that SCEAT3D with merging mean absolute error with subtracting difference for the vertices of objects (MAE-) is a suitable descriptor and suitable scheme for encrypting huge 3D animated textured objects.