Authors :

Ravi Patel

Address :

Assistant Professor, Department of Machine Learning, Indian Institute of Technology, Bombay, India

Abstract :

Securing digital content from unauthorized access, modification, misuse, or loss is paramount, which is why media security must be ensured. As the need for encrypted communication grows, there is a significant demand for compression and encryption methods that maintain image quality and allow for scalable decoding. Traditional encryption techniques frequently need more bandwidth and storage space and don't necessarily offer the optimum compression. This study introduces SEIC-MBTRGNN, a novel integrated approach to address these concerns. It stands for Scalable Encrypted Image Coding via Modified Block Truncation and Recursive Graph Neural Networks. The system includes a Recursive Graph Neural Network (RGNN), Scalable Coding of Encrypted Images (SCEI), and Modified Absolute Moment Block Truncation Code (MAMBTC). In the first stage, MAMBTC compresses the image data by preserving the crucial elements. Pseudo Random Number Generator (PRNG) encryption is applied to compressed image for further security. With RGNNs, operational flexibility and security are improved for scale-coded graph-structured data. After decryption, the PRNG uses MAMBTC to resize and recreate the image by restoring compressed pixels. Following this, bilinear interpolation is used to rebuild the initial image. The performance and security of this technology are 30% higher than those of state-of-the-art compression and encryption algorithms, according to the experimental results. With an accuracy rate of 95%, the RGNN layer provides an upgraded, versatile, and effective way for secure picture processing in modern media security systems. The method's scalability allows processing to be tuned to meet resource availability or unique security needs. This revolutionary approach provides a comprehensive answer to the problem of media security by balancing robust encryption with efficient compression while maintaining picture quality.

Keywords :

Encrypted Images, Scalable Coding, Recursive Graph Neural Networks, Media Security, Image Compression, Modified Absolute Moment Block Truncation Code