SELECTIVE ENCRYPTION OF VIDEO INFORMATION BASED ON SEMANTIC SEGMENTATION USING U-NET NEURAL NETWORK

In the digital age, video information has taken a leading place among data types in terms of volume and significance. Large amounts of visual data are created daily using video cameras, mobile gadgets, drones and network services, and a significant part of this content may contain personal or confidential information. Although traditional full encryption of the video stream guarantees a high level of protection, it is accompanied by a number of disadvantages: high load on computing resources, delays during data transmission and difficulties in implementing real-time processing. The paper solves the scientific and applied problem of selective cryptographic protection of video information based on automatic object identification using the U-Net neural network. A multi-round encryption method is proposed, which implements global permutation of pixels and byte-by-byte replacement of their values in the selected image area. The purpose of the research is to develop an effective method for cryptographic protection of target image areas after their identification, with an emphasis on diffusion, confusion, and protection from statistical analysis. A system of coupled chaotic tent mappings is used to generate the key material, and the PBKDF2 mechanism with a cryptographic salt is used for parameterization. The paper also considers the use of XOR and modular addition operations, which alternate between rounds, which increases the cipher's resistance to attacks. An experimental analysis of the results is carried out, which confirms the effectiveness and resistance of the developed method to changes in the input data and key information. The result of the work was the development of a new 9-round symmetric encryption algorithm for selected ROIs. This algorithm converts the user-supplied password and a unique random salt into all the necessary key material using a standard key strengthening function. Each of the nine encryption rounds includes two main operations: a global permutation of all pixels in the ROI and a byte-by-byte replacement of their values. The replacement operation uses key streams generated by a system of coupled chaotic mappings, the settings of which are individual for each round. The type of replacement operation alternates between rounds. To verify the integrity and authenticity of the encrypted data, a standard message authentication code mechanism is used.