Scene classification is one of the areas of remote sensing image processing that is gaining much attention. Aiming to solve the problem of the limited precision of optical scene classification caused by complex spatial patterns, a high similarity between classes, and a high diversity of classes, a feature cross-layer interaction hybrid algorithm for optical remote sensing scene classification is proposed in this paper. Firstly, a number of features are extracted from two branches, a vision transformer branch and a Res2Net branch, to strengthen the feature extraction capability of the strategy. A novel interactive attention technique is proposed, with the goal of focusing on the strong correlation between the two-branch features, to fully use the complementing advantages of the feature information. The retrieved feature data are further refined and merged. The combined characteristics are then employed for classification. The experiments were conducted by using three open-source remote sensing datasets to validate the feasibility of the proposed method, which performed better in scene classification tasks than other methods.

The growing scope of image degradation encountered in remote sensing detection has sparked significant interest in the application of deep learning methods. To address the challenges posed by high-frequency signal loss and structural distortion in reconstructed remote sensing images, a transformer-based skip network with frequency attention (named FATSNet) is proposed to improve the model's ability. The model comprises two key blocks: encoder and decoder blocks, which incorporate two novel frequency attention modules. In the encoder block, a discrete wavelet transform (DWT) module is designed to implement the representation of detailed abstract content features. The decoder block utilizes a frequency-aware transformer (FAT) module based on fast Fourier transform (FFT) to enhance the clarity of the image structures and boundaries. To improve the robustness of the network during feature fusion, a self-adapting weighting module (SAWM) is proposed by dynamically adjusting the weights of feature channels to make more effective use of feature information. The experimental results with public datasets demonstrate the superiority of the FATSNet model over other leading methods. Additionally, a case study of the detection of Lake Buridun was used to verify the practicality of the proposed model.

In this paper, a new image analysis method based on an in-situ microscopic imaging system is proposed for detecting micro crystals in cooling crystallization. Due to the limitation of measurement technology, it is a challenge to extract the evolutionary information of micro crystals, which are too small to be precisely analyzed by in-situ images, e.g. crystals at the initial crystallization stage. An improved deep-learning model is used to enhance the image resolution of micro crystals, thus more effectively obtaining the crystal shape and size information. In addition, a valid size calibration method by simulating particle motion is proposed. Consequently, image size measurement can be easily performed for crystals by using an axis-based algorithm. Experimental verifications on β-form L-glutamic acid crystallization were performed to demonstrate the effectiveness of the proposed method for detecting micro crystal information.

In this paper, a novel in-situ measurement method with stereoscopic image analysis is proposed for monitoring crystal length and width distributions during a cooling crystallization process, based on a binocular telecentric imaging system. Firstly, a stereoscopic imaging calibration model is established for using binocular telecentric cameras to conduct in-situ measurement during a cooling crystallization process. Secondly, an enhanced algorithm is presented to improve matching crystal images from binocular image pairs, based on in-situ image preprocessing and segmentation. Thirdly, four key corners related to crystal length and width are determined by using the boundary features of matched crystal image projections. Finally, the length and width of each crystal are computed based on the reconstructed key corners in three-dimensional (3D) coordinate space, by using the established stereoscopic calibration model. Experimental validation on the proposed stereoscopic imaging calibration model and 3D reconstruction of key corners is carried out via in-situ measurement of a microscale checkerboard plate inserted into a cooling crystallizer. In-situ measurements of crystal length and width distributions during the cooling crystallization process of β-form L-glutamic acid are conducted to verify the effectiveness and advantage of the proposed measurement method.

In this paper, a crystal image analysis method is presented to measure and predict crystal sizes, based on cooling crystallization of β-form L-glutamic acid (LGA) by using an in-situ non-invasive imaging system. The proposed method consists of image restoration, image segmentation, crystal size measurement, and size prediction. To cope with the effects of noise pollution, uneven illumination and movement blurring, the image processing method is conducted for segmenting crystal images captured from the stirring reactor. Thus, the crystal size distribution for crystal population is obtained by using a probability density function. In addition, a short-term prediction method is given for crystal sizes. An experimental study on the cooling crystallization process of β-form LGA is shown to demonstrate the effectiveness of the proposed method.

Fault diagnosis for Autonomous underwater Vehicle (AUVs) is a key technology for ensuring the safety of AUVs and an important skill for enabling them to autonomously perform tasks underwater for long periods. The effectiveness of current diagnostic methods is affected by the reliability of expert knowledge and the accuracy of model establishment. In addition, some data-driven diagnostic methods lack robustness. Unlike traditional model-based fault diagnosis methods, this paper proposes a fault diagnosis method for AUVs based on the LSTM (Long Short-Term Memory) algorithm. LSTM is good at processing time series data and can learn complex temporal patterns. Therefore, the LSTM model is used to learn the mapping of state data to its corresponding fault types. The underwater environment in which AUVs work is complex and ever-changing, and packet loss may occur during data transmission, resulting in partial loss of online data. To address this issue, this paper fills in missing values during the feature processing stage and then uses a BiLSTM-Attention-MiniLoss algorithm to enhance the robustness of the diagnostic model. Finally, the fault diagnosis accuracy of the original LSTM and the BiLSTM-Attention-MiniLoss was compared based on an open-source dataset under different degrees of data loss. The experimental results showed that the fault diagnosis methods for AUV based on LSTM and the BiLSTM-Attention-MiniLoss could predict the type of fault based on the navigation status data of the AUV, with BiLSTM-Attention-MiniLoss performing better.