An image compressed sensing algorithm based on adaptive nonlinear network*

A novel ghost imaging-based optical cryptosystem for multiple images using the integral property of the Fourier transform is proposed. Different from other multiple-image encryption schemes, we mainly construct the modulation patterns related to the plaintext images to realize the encrypted transmission of multiple images. In encryption process, the first image is encrypted by the ghost imaging encryption scheme, and the intensity sequence obtained by the bucket detector is used as the ciphertext. Then modulation patterns of other images are constructed by using the integral property of the Fourier transform and used as the keys. Finally, the ciphertext and keys are transmitted to the receiver to complete the encryption process. During decryption, the receiver uses different keys to decrypt the ciphertext and gets different plaintext images, and decrypted images have no image aliasing problem. Experiments and simulations verify the feasibility, security, and robustness of the proposed scheme. This scheme has high scalability and broad application prospect, which provides a new idea for optical information encryption.

“…The development of machine learning also makes it possible to reduce the amount of transmitted data and speed up information reconstruction. [31,32]…”

Section: Discussionmentioning

confidence: 99%

Ghost imaging-based optical cryptosystem for multiple images using integral property of the Fourier transform*

Yi¹,

Zhang²,

Ye³

et al. 2021

“…Convolutional neural networks (CNNs) have been successfully applied in several fields, such as image processing, object detection, and machine vision, with promising prospects in aerospace. [1][2][3][4] The Xilinx Zynq-7020 is an allprogrammable system-on-chip (SoC) that features a dualcore ARM Cortex-A9 processor and Xilinx 7-series field programmable gate array. Due to its configurability, design flexibility, and high performance-to-power ratio, this SoC has become a significant application direction for developing the CNN system.…”

Section: Introductionmentioning

confidence: 99%

Single event effects evaluation on convolution neural network in Xilinx 28 nm system on chip

Zhao 赵,

Du 杜,

Xiong 熊

et al. 2024

Convolutional neural networks (CNN) have great exhibit excellent performance in the areas of image recognition and object detection, which can enhance the intelligence level of spacecraft. However, in aerospace, energetic particles, such as heavy ion, proton and alpha particle, can induce single event effects (SEE) that lead to malfunction of CNN, and significantly impact the reliability of CNN system. In this paper, the MNIST CNN system was constructed based on a 28 nm System-on-Chip (SoC), and then the alpha particle irradiation experiment and fault injection were applied to evaluate the SEE of the CNN system. Various types of soft errors in the CNN system have been detected, and the SEE cross sections have been calculated. Furthermore, the mechanisms behind some soft errors have been explained. The research will provide technical support for designing of radiation-resistant artificial intelligence chips.

“…Based on the previous discussion of image encryption algorithms based on CS, it is worth noting that while some algorithms have used CS to achieve efficient image compression, the trade-off is reduced image quality and potential security vulnerabilities in the encryption process. [32][33][34] This has led to a growing interest in designing image encryption algorithms that balance robustness and security. One approach that has gained attention is the use of optical encryption methods, such as the double random-phase encoding algorithm.…”

Section: Introductionmentioning

confidence: 99%

Optical image encryption algorithm based on a new four-dimensional memristive hyperchaotic system and compressed sensing

Long

Jiang

et al. 2023

Some existing image encryption schemes use simple low-dimensional chaotic systems, which makes the algorithms insecure and vulnerable to brute force attacks and cracking. Some algorithms have issues such as weak correlation with plaintext images, poor image reconstruction quality, and low efficiency in transmission and storage. To solve these issues, this paper proposes an optical image encryption algorithm based on a new four-dimensional memristive hyperchaotic system (4-D MHS) and compressed sensing (CS). Firstly, this paper proposes a new 4-D MHS, which has larger key space, richer dynamic behavior and more complex hyperchaotic characteristics. The introduction of CS can reduce the image size and the transmission burden of hardware devices. The introduction of double random phase encoding (DRPE) enables this algorithm has the ability of parallel data processing and multi-dimensional coding space, and the hyperchaotic characteristics of 4-D MHS make up for the nonlinear deficiency of DRPE. Secondly, a construction method of the deterministic chaotic measurement matrix (DCMM) is proposed. Using DCMM can not only save a lot of transmission bandwidth and storage space, but also ensure good quality of reconstructed images. Thirdly, the confusion method and diffusion method proposed are related to plaintext images, which require both four hyperchaotic sequences of 4-D MHS and row and column keys based on plaintext images. The generation process of hyperchaotic sequences is closely related to the hash value of plaintext images. Therefore, this algorithm has high sensitivity to plaintext images. The experimental testing and comparative analysis results show that proposed algorithm has good security and effectiveness.