Matrix Models of Cryptographic Transformations of Video Images Transmitted From Aerial-Mobile Robotic Systems

Introduction

In various types of industrial activity of a person, as well as in his daily life, photo, static and dynamic images of various formats, video information about various surrounding objects are widely used today. A characteristic feature of modern digital video surveillance systems that are used in unmanned aircraft, when analyzing the traffic situation, remote monitoring and demonstration of emergency or other situations, in security activities, recording events in places of public events, is their distribution. Video information transmitted in such systems, although it is not always secret and has a small period of actuality, is often undesirable for mass distribution and use. In the above-mentioned video systems, especially, such as security, telemedicine and special-purpose systems, in intelligent robotic complexes, not only the tasks of perception, accumulation and transmission of digital video images, but also their protection from unauthorized access, problems of distortion, substitution of information and verification of the integrity of video files are actual tasks. Transmission of video information over open communication channels, IP-networks, and widespread use of wireless technologies for these video systems makes it possible to access information to unauthorized users. The above-mentioned tasks are of particular relevance for mobile robotic and distributed systems implemented on the basis of embedded-class IP modules, for which there are limitations on the computation speed and free computational resource. The specificity of the above systems is that in most cases the transmitted video information is relevant for a short period of time and the use of complex well-studied and widely used cryptographic methods of protection, and especially those requiring significant computational resources, is not required. The analysis showed that in embedded class systems, which include IP-modules of distributed or airmobile video systems, standard cryptographic algorithms are limited, and more often, simpler cryptographic primitives and masking methods are used. The masking information is meant the process of converting digital visual information to a noise-like view in order to protect against unauthorized access, and unmasking is the process of reversely converting masked visual information into restored (outgoing) by applying operations that are inverse to the direct masking procedures. Masking transformations are one of the alternatives to cryptographic methods of photo and video information protection. In the authors' opinion, masking is a special case of some transformations, which are not always cryptographic standard ones. Besides, it is necessary to distinguish matrix masking, as transformation processes using matrixes and matrix procedures follow only when matrices and operations on them appear in the corresponding models. In some cases, by cryptographic masking, authors imply direct and inverse image transformations in which elements of cryptographic methods are used, and the result of masking is the destruction of images to a form that is visually perceived as noise. We stand on the view that reliable and cryptographic protection requires cryptographic procedures and matrix models that transform the original video image into not only noise, but also to provide some important entropy characteristics. A feature of entropy, as a generalized concept of measurement of the uncertainty of processes, is the fact that it reduces to some numerical values, which can be operated as a relative value, and to characterize with it the quality of cryptographic transformations or masking. The entropy of the uniform distribution law (white noise) is an idealization that is maximized and has the greatest disinformation action. The use of masking as a method of protecting video information with a short time of relevance is associated with solving the problems of generating masked data structures, their presentation, exchange between the receiver and the transmitter, and storing and unmasking information. Often, masking takes into account the specific structure of video frames (photos), algorithms for their compression and transmission protocols. It is known that frames are represented as matrix arrays of pixels, the values (intensities) of elements of which are displayed by digital codes, and therefore the matrix apparatus and the operations for converting them are natural. The main types of images that need to be perceived, cryptographically transformed and transmitted in intellectual robotic video surveillance subsystems are half-tone, binary and full-color images, although many more and more hyper-spectral images are being used. The main format of digital images, which directly stores the values of pixel intensities of images obtained from the video-matrix is a Bitmap Picture (BMP), providing storage of images of various sizes and depths. Matrix operations and the matrices themselves are widely used for mathematical modeling of various processes and systems. Matrices are the basic apparatus for most engineering and scientific calculations. Computations over matrices, although laborious, are focused on parallel computing, on significant increases in computational performance, and are a classic example and direction for the further development of more intelligent computer architectures of parallel action. The emerged multi-core processors, graphics accelerators, digital signal processing (DSP) processors, structures on the FPGA, essentially support vector and matrix calculations and increase the speed and performance of the latters. Therefore, it is precisely for more modern hardware implementations that matrix models are ideally suited for implementations of cryptographic or similar methods for transforming and masking information objects in order to protect them during transmission. Matrix algebra and its operations are well studied, they are structured, regular, and easily mapped to hardware matrix structures that provide parallelization of computations, increasing computational performance and efficiency. In addition, such structures are more efficiently implemented using DSP or FPGA, which is important for systems of embedded classes, especially small and mobile ones. The actual task is to create such cryptographic procedures for direct and inverse transformation of video information, which use fully matrix models and procedures that are easily mapped to the corresponding matrix equipment. At the same time ensuring their simplicity, meeting the requirements of speed and computation performance while ensuring the best entropy characteristics .