Scholarlink

<p>Hyperspectral images are crucial for remote sensing, military applications, medical imaging, and satellite navigation systems. Maintaining the confidentiality, privacy, security, and intellectual property rights of the HSIs is essential. However, generalized traditional encryption methods are less secure and robust due to larger spectral bands, high variability in spectral bands, redundancy in spectral bands, and higher encryption time. This article presents selective hyperspectral image encryption based on a deep learning-based hyperchaotic image encryption algorithm that improves the security and robustness of the hyperspectral images. Here, a deep neural network is utilized to enhance the randomness of the hyperchaotic sequence key and improve encryption. The performance of the deep learning-based hyperchaotic image encryption is estimated on the PAVIA, SALINAS, Indian Pines, Kenndy Space Center, and Cuprite datasets based on Mean Square Error, Peak signal-to-noise ratio, Structural Similarity Index metrics, Correlation Coefficient, Number of Pixel Change Rate, Unified Average Changing Intensity and entropy. The proposed method offers the MSE in the range of 1.946 to 2.007, PSNR in the range of 45.10 dB to 46.31 dB, SSIM in the range of 0.953 to 0.982, CC in the range of 0.02 to 0.015, NPCR in the range of 98.34, UACI in the range of 27.59 for five sample hyperspectral images. The proposed scheme provides improved security and robustness against distinct encryption attacks such as Brute Force Attack and chosen-plaintext-attack. It offers superior entropy of 7.85 for brute force attack and 7.83 for the chosen-plaintext attack.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhanced selective hyper chaotic encryption using deep neural network for hyperspectral images

  • 期刊论文

  • 发表日期：

    2025年1月10日

查看全文
  • Minal Bodke,
  • Sangita Chaudhari

摘要

Hyperspectral images are crucial for remote sensing, military applications, medical imaging, and satellite navigation systems. Maintaining the confidentiality, privacy, security, and intellectual property rights of the HSIs is essential. However, generalized traditional encryption methods are less secure and robust due to larger spectral bands, high variability in spectral bands, redundancy in spectral bands, and higher encryption time. This article presents selective hyperspectral image encryption based on a deep learning-based hyperchaotic image encryption algorithm that improves the security and robustness of the hyperspectral images. Here, a deep neural network is utilized to enhance the randomness of the hyperchaotic sequence key and improve encryption. The performance of the deep learning-based hyperchaotic image encryption is estimated on the PAVIA, SALINAS, Indian Pines, Kenndy Space Center, and Cuprite datasets based on Mean Square Error, Peak signal-to-noise ratio, Structural Similarity Index metrics, Correlation Coefficient, Number of Pixel Change Rate, Unified Average Changing Intensity and entropy. The proposed method offers the MSE in the range of 1.946 to 2.007, PSNR in the range of 45.10 dB to 46.31 dB, SSIM in the range of 0.953 to 0.982, CC in the range of 0.02 to 0.015, NPCR in the range of 98.34, UACI in the range of 27.59 for five sample hyperspectral images. The proposed scheme provides improved security and robustness against distinct encryption attacks such as Brute Force Attack and chosen-plaintext-attack. It offers superior entropy of 7.85 for brute force attack and 7.83 for the chosen-plaintext attack.