Aiming at the problems of small key space, low security of encryption structure, and easy to crack existing image encryption algorithms combining chaotic system and DNA sequence, this paper proposes an image encryption algorithm based on a hidden attractor chaotic system and shuffling algorithm. Firstly, the chaotic sequence generated by the hidden attractor chaotic system is used to encrypt the image. The shuffling algorithm is used to scramble the image, and finally, the DNA sequence operation is used to diffuse the pixel value of the image. Experimental results show that the key space of the scheme reaches 2327 and is very sensitive to keys. The histogram of encrypted images is evenly distributed. The correlation coefficient of adjacent pixels is close to 0. The entropy values of encrypted images are all close to eight and the unified average change intensity (UACI) value and number of pixel changing rate (NPCR) value are close to ideal values. All-white and all-black image experiments meet the requirements. Experimental results show that the encryption scheme in this paper can effectively resist exhaustive attacks, statistical attacks, differential cryptanalysis, known plaintext and selected plaintext attacks, and noise attacks. The above research results show that the system has better encryption performance, and the proposed scheme is useful and practical in communication and can be applied to the field of image encryption.
Multiple-source DOA (Direction of Arrival) estimation is vital for array processing, radar, and integrated sensing and communications, yet classical subspace methods degrade under low signal-to-noise ratios and snapshot-starved conditions due to inaccurate sample covariance matrices. To address this, we propose a Toeplitz-enhanced neural network (TENN-DOA) for DOA estimation, a hybrid physics-informed framework for uniform linear arrays that combines the array signal processing prior with a lightweight learning-based regressor. The front end explicitly enforces the Hermitian–Toeplitz structure and fuses the projected matrix with the sample covariance via an analytically derived optimal shrinkage coefficient, yielding a robust covariance estimate. This enhanced representation is mapped onto an overcomplete angular dictionary, producing a feature sequence structurally coupled with the array manifold. A pooling-free one-dimensional convolutional neural network with decreasing kernel sizes starts with large kernels to capture the broad spectral envelope from grid mismatch, and then regresses to a pseudo spatial spectrum under multi-hot supervision for grid-point estimates. The Monte Carlo simulation results show that under the conditions of low signal-to-noise ratio, limited snapshots and the simulated ideal uniform linear array scenario, TENN-DOA achieves a higher resolution probability and lower root mean square error compared with MUSIC, TLS-ESPRIT and the deep learning-based baseline algorithm DA-MUSIC.