This paper presents a fast SAR imagery algorithm for Ground Moving Target Imaging (GMTIm) based on the slope detection strategy combined with Time-Frequency Representation (TFR), which is known as Lv’s Distribution (LVD). This fast imagery algorithm focuses on the solution of the ambiguity problems and relevant heavy computing load in SAR imagery. Firstly, according to the relationship between the slope of the range walk trajectory and the cross-track velocity of moving target, a new high-efficiency slope detection strategy based on gradient and level-line angle is presented in the image domain. Then, the Doppler centroid shift induced by cross-track velocity can also be obtained. Secondly, owing to the cross-track velocity estimated before, the Range Walk Migration Correction (RWMC) can be performed to concentrate the echo response of the moving target into a single range cell. Finally, due to the superior performance in representing multi-component Linear Frequency Modulation (LFM) signal, LVD is adopted here to represent the Doppler chirp rate of multiple moving targets in a Doppler Centroid Frequency and Chirp Rate domain (CFCR). The performance of the proposed algorithm is evaluated in terms of superiority and effectiveness using simulations, and the comparison between the proposed algorithm and the other conventional algorithms is also presented.

Traditional near-field synthetic aperture radar (SAR) imaging algorithms reveal target features by exploiting signal amplitude and phase information. However, electromagnetic wave propagation is constrained by short distance. Therefore, the spherical wave approximation needs to be considered. In addition, it is also limited by equipment ambient noise, azimuth-distance coupling, wave scattering, and transmission power. Both the amplitude and phase of the signal suffer from the interference of multiple clutter, so they cannot be effectively utilized. To address these issues, this paper introduces a covering penetration detection system based on an improved Range Migration Algorithm (IMRMA) imaging method. Firstly, the proposed method minimizes interferences from the front end of the system using an optimized window to balance denoising and information preservation. Next, interval non-uniform interpolation, instead of Stolt interpolation decoupling, is employed to reduce the computational overhead significantly. To minimize the effects due to wave scattering and propagation loss, distance information is enhanced using amplitude and phase compensation. This reduces scattering effects and enhances image quality. An experimental system is constructed based on a vector network analyzer (VNA) to image the target. The proposed method takes about half the time of traditional RMA. The PSNR in the chunky bowl experiment is higher than 14 dB, which is higher than all the compared methods in the paper. The test results show that the designed system and the reported method can effectively achieve high-resolution images by strengthening the target intensity and suppressing the environmental artifacts.

Molecular property prediction is essential for drug screening and reducing the cost of drug discovery. Current approaches combined with deep learning for drug prediction have proven their viability. Based on the previous deep learning networks, we propose the Molecular Information Fusion Neural Network (MIFNN). The features of MIFNN are as follows: (1) we extracted directed molecular information using 1D-CNN and the Morgan fingerprint using 2D-CNN to obtain more comprehensive feature information; (2) we fused two molecular features from one-dimensional and two-dimensional space, and we used the directed message-passing method to reduce the repeated collection of information and improve efficiency; (3) we used a bidirectional long short-term memory and attention module to adjust the molecular feature information and improve classification accuracy; (4) we used the particle swarm optimization algorithm to improve the traditional support vector machine. We tested the performance of the model on eight publicly available datasets. In addition to comparing the overall classification capability with the baseline model, we conducted a series of ablation experiments to verify the optimization of different modules in the model. Compared with the baseline model, our model achieved a maximum improvement of 14% on the ToxCast dataset. The performance was very stable on most datasets. On the basis of the current experimental results, MIFNN performed better than previous models on the datasets applied in this paper.