Microsphere-assisted microscopy (MAM) is a technique aimed at enhancing the lateral resolution of optical microscopy, enabling high lateral resolution profile measurement when combined with interferometry. MAM can operate in lift mode, facilitating the selection of regions of interest and expanding the field of view. The analysis of the lifting mode of microspheres in microsphere-assisted interferometry is still insufficient, which affects the longitudinal measurement accuracy of microsphere-assisted interferometry. The phase transmission mechanism of the microsphere was simulated in this paper, and the relationship between the phase distribution below the microsphere and the distance between the microsphere and the sample was summarized. A combined system of microsphere-assisted white light interference microscope was constructed, and the magnification factor and phase distribution of the microsphere in lift mode was measured through atomic force microscope atomic force microscope (AFM) control of the microsphere’s position. The experiment validated the simulated results of microsphere phase transmission, providing a theoretical foundation for microsphere-assisted interferometry(MAI) in lift mode.
Underwater skylight polarization images can be utilized for heading measurement, offering advantages such as anti-interference and no accumulation error. A method for underwater polarization image heading estimation was presented based on the superposition of light vectors in the zenith region, with the aim of obtaining the heading angle in underwater environments. The polarization information from the zenith region was analyzed and extracted in conjunction with attitude information. Subsequently, the underwater refracted light polarization vector was synthesized with the atmospheric polarization light vector within the zenith region using superposition operations. Through optimization of the polarization vector, the heading angle of underwater targets was accurately determined. Experimental results demonstrated that the root mean square error (RMSE) of the heading angle in the proposed method was 0.30° in the tank experiment and 0.41° in the marine experiment. Moreover, in the oceans at depths of 0.98 m, 4.89 m, and 5.94 m, the RMSE of the solar azimuth was 1.10°, 2.03°, and 3.04°, respectively.
The traditional tongue diagnosis process has the problem of poor objectivity. Applying computer vision technology to tongue diagnosis can effectively promote its objectivity. Binocular stereo vision combined with structured light fringe projection technology is a common method for 3D measurement. However, in the measurement scenario of tongue diagnosis, due to the presence of saliva and fluids on the tongue surface, there are high-reflectance areas with significant random distribution in the fringe images, leading to errors in phase calculation and point cloud loss. A trinocular measurement system was proposed based on fringe projection, where a trinocular system and three binocular subsystems were composed of three cameras. Dual-epipolar constraint based on phase and order constraints was introduced to enhance the accuracy of trinocular stereo matching. Supplementary matching points were utilized to optimize the trinocular matching point sets, reconstructing point clouds in high-reflectance areas. The results indicated that, compared to traditional binocular systems, this system achieved improved matching and reconstruction accuracy. Particularly in real tongue surface measurements, it could generate point clouds with clear textures and complete features. It could effectively measure the highly reflective area of the tongue surface and facilitate objective tongue diagnosis.
Aiming at the problems of poor contrast of fusion results, blurring of target margins and loss of background detail information under low illumination conditions of traditional infrared and visible image fusion algorithms, an infrared and visible image fusion algorithm based on multi-scale decomposition with Retinex enhancement was proposed. Firstly, the single-scale Retinex(SSR) algorithm information enhancement process was performed on the weak visible image using Retinex. Secondly, the source image was multi-scale decomposed using cross bilateral filtering to successively obtain the image information of the base layer image and the detail layer, and the fusion method combining the absolute value maximization strategy and guided filtering was used for the base layer image, and the fusion method of constructing weight map and significant map was used for the detail layer image. Finally, the processed base layer image and detail layer image were weighted to obtain the fused image. From the subjective analysis, the proposed method could effectively extract and fuse the important information in the source image, and obtain the image with high fusion quality and natural and clear visual effect. From the objective evaluation, the average accuracy of the proposed method on AG, SF, CE, and FMI was optimal when compared quantitatively with various fusion results.
Aiming at the lack of semantic correlation between the parameters of expectation maximization attention(EMA) algorithm and images and the lack of attention to inter-channel information, a dual attention network EMA+ algorithm was proposed. Two modules were designed: spatial attention module and channel attention module. The EMA algorithm was used as the main structure by the spatial attention module. In the responsibility estimation step, the feature map itself was used as the initial parameter in the expectation maximization(EM) algorithm, and the semantic association between the parameter and the feature map was increased. Efficient channel attention(ECA) was used in the channel attention module by using one-dimensional convolution to learn the interactive information between channels. It avoided breaking the direct correspondence between channels and their weights due to dimensionality reduction operations. EMA+ significantly improved semantic segmentation tasks’ performance by fusing spatial attention modules and channel attention modules. The experimental results showed that EMA+ has achieved better intersection-over-union than EMANet and other methods on PASCAL VOC 2012 and some more complex datasets, and had better generalization ability.
Edge detection is a fundamental method in image processing and computer vision. Aiming to address the issues of roughness and blurriness in edges generated by deep learning-based edge detection technology, a refined edge detection(RED) model based on richer convolutional features(RCF) for edge detection was proposed. In this model, RCF was used as the baseline network. Some downsampling operations in the backbone network were removed, and the coordinate attention(CA) module and hybrid dilated convolution were added to the backbone network. The number and parameters of the compression layers were changed in the deep supervision module, and smooth compression for reducing feature dimensionality was adopted. In the final fusion module, a cross-layer cross-fusion module was used to fuse the information from high and low layers. The RED model was trained and tested on the extended BSDS500 dataset. The optimal dataset scale(ODS) and the optimal image scale(OIS) of the dataset were 0.809 and 0.832, respectively, as evaluated on the BSDS500 benchmark. The experimental results showed that RED model extracted clearer and more detailed edge contours, and the extracted edge information was more comprehensive and abundant.
A hybrid forecasting model was proposed to improve the accuracy of short-term photovoltaic (PV) power generation forecasting, which combined the clustering of trained self-organizing map(SOM) network and optimized kernel extreme learning machine(KELM) method. First, a pure SOM was employed to complete the initial partitions of the training data set. Then clustering was executed on the trained SOM network by fuzzy C-means(FCM). Meanwhile, the davies-bouldin index(DBI) was hired to determine the optimal size of clusters. Finally, in each data partition, the regional KELM model was built with the KELM optimized by differential evolution, or the regional linear regression(MR) model was built with the multiple MR using the least square method to complete the coefficient evaluation. In addition, varying local multiple regression model was also proposed based on SOM. The proposed model based on SOM-FCM and KELM was employed to one-hour-ahead PV power forecasting instances of three different solar power plants provided by the GEFCom2014. Compared with other control models, the mean absolute error (MAE) of plant 1 was reduced by 61.41%, that of plant 2 by 60.19%, and that of plant 3 by 58.92%. The root means square errors (RMSE) of plant 1 was reduced by 52.06%, that of plant 2 by 54.56%, and that of plant 3 by 51.43% on average. The forecasting accuracy was significantly improved with the proposed model.
Efficient recovery of the autonomous underwater vehicle (AUV) in complex underwater environments is a major technological challenge. A terminal docking method using monocular vision was presented to detect and track a circular light source marker at the docking station. To improve the recognition of the light source feature, the Canny edge detection algorithm was improved, the adaptive threshold method was used to dynamically adjust the light source contour, and the minimum enclosing circle method was used to determine the light source center when the threshold was optimal. In addition, geometric position and area constraints were used to eliminate interference from water surface reflections. For visual localization and tracking, Zhang’s calibration method was used to obtain the internal and distortion parameters of the camera, yaw and pitch errors of the AUV were estimated by comparing the light source center with the camera image center, then the position-attitude PID controller was used to achieve rapid attitude adjustment. The pool experimental results showed that the approach was simple, practical, and robust, providing a technical reference for future reliable recovery of underwater robots.
In order to improve the heating efficiency of the electric heating system for heavy oil wells and solve the current situation that the heating operation control of the medium frequency power supply in the oilfield mainly relies on manual experience settings and low-efficiency automatic control resulting in electric power waste, an embedded control device for heavy oil electric heating was designed. Based on the non-embedded oil well production electrical signal, the electric power of the heavy oil production motor was used as the closed-loop feedback signal to obtain the reference input of the optimal oil temperature. On this basis, through the differential feedback link of the wellhead oil temperature, the temperature hysteresis effect was improved, and the energy-saving optimization control of the medium frequency power supply was realized. Through the design of the embedded hardware core circuit module and the development of the main software functions, not only the dynamic control of the electric heating process of heavy oil wells was realized, but also realized the real-time monitoring of various operating indicators on-site and in the cloud. The heavy oil well field test analysis showed that the designed heavy oil electric heating embedded control device could meet the production requirements of energy saving and safety in the well field, and the energy consumption was saved by 20%.
Aiming at the problem that traditional vehicle trajectory matching algorithms based on hidden Markov model(HMM) cannot have both accuracy and time efficiency in complex and special road sections, a vehicle trajectory matching method based on improved HMM modeling was proposed. In the determination of candidate road sections, grid index was generated to improve the overall retrieval efficiency. The improved HMM model integrated heading angle factors in the calculation of launch probability, considered the deviation effect caused by vehicle speed on heading angle, and set empirical factors for adjustment. At the same time, considering the factors such as the excessive error of the observation value before and after and the curve section, the actual travel distance of the vehicle within the unit sampling interval was used instead of the observation distance value to ensure the accuracy of the calculation of the transfer probability. Finally, the measured data was used to conduct experiments to verify the performance of the improved algorithm. The experimental results indicated that the matching accuracy of this method was about 94.0%, which was 2.8% higher than that of the traditional HMM trajectory matching method. It also had certain advantages in improving time efficiency and matching accuracy of complex road sections. The single-point matching time was reduced by about 0.9 ms, suitable for matching under complex road conditions such as intersections, overpasses, and parallel sections.
The photomultiplier tube (PMT) is an important device for micro-light detection, and the detection of light intensity using photon counting method can significantly reduce the interference of noise, but the sensitivity, gain, and dark noise of PMT cathode are easily affected by the ambient temperature, which leads to the instability of the output pulse amplitude and affects the detection performance of the system for micro-light. A high voltage gain compensation and threshold correction system was designed for PMT through microcontroller unit (MCU), digital to analog converter (DAC), and other modules. A hybrid function model of temperature and high voltage compensation increment Vh and threshold compensation increment Vt was constructed by analyzing the cathode saturation current, sensitivity, and dark noise of PMT at different temperatures. And the compensation increment hybrid model was used to compensate the PMT counts output for adaptive temperature drift compensation. Experiments using this method with Hamamatsu’s end-window PMT CR135 demonstrated that this system had a good output signal-to-noise ratio for large temperature variations, with an average count rate improvement of 0.19 at -20 ℃. Even though a small amount of dark noise was introduced, the detection performance was substantially improved.
The electromagnet’s state of operation plays a crucial role in maintaining the operational mechanism of circuit breakers, as it acts as a trigger for the opening and closing action of the mechanism. However, due to the interference of complex electrical environments and the limitation of sensing conditions, there are significant deficiencies in the robust condition monitoring of electromagnets. A hybrid two-stage method was proposed to diagnose the running state of the electromagnet of the operating mechanism. By intelligently identifying the key characteristic points of the electromagnet current signal, the proposed method indirectly realized the intelligent diagnosis of the state of the electromagnet. In the first identification stage, an intelligent U-Net neural network suitable for the one-dimensional signal was proposed to realize the adaptive identification of crucial feature points via the obtained current signal of electromagnets. In the second condition monitoring stage, based on the position and the current value of the key feature points, the operating state of the electromagnet could be identified specifically. The experimental findings demonstrated that the suggested strategy was capable of successfully identifying the key characteristic points, with a near-perfect recognition success rate. The proposed method realized the adaptive identification of various electromagnet faults with only a few fault samples, which provided a guarantee for robust state identification of electromagnets and had the advantage of high interference resistance.
Rolling bearing is a critical component in the rotating machinery, which directly affects the reliability of the equipment. The artificial intelligence-enabled bearing fault diagnosis model has achieved impressive successes over the years. However, rolling bearings’ imbalanced data sets (normal samples are much larger than failure samples) degrade the diagnostic performance. To address this issue, a bidirectional generative adversarial network(BiGAN) based fault diagnosis method was proposed. First, the signal was denoised via the ensemble empirical mode decomposition(EEMD) to automatically distribute it to a suitable reference scale and avoid modal aliasing. Then, the BiGAN model with gradient penalty term was constructed to expand the fault samples, where the min-max normalization was included. Finally, based on the enhanced training set, the convolutional neural network was established with batch normalization and maximum pooling layers. Experimental results proved that the proposed method improved fault diagnosis accuracy and robustness.
Aiming at the sparsity of point cloud data and the low accuracy of spatial alignment exhibited by millimeter-wave frequency-modulated continuous-wave (FMCW) radar in outdoor motion scenarios, a lightweight model for spatial alignment was proposed. This method was specifically tailored for point cloud processing across consecutive multi-frames in outdoor motion scenes captured by millimeter-wave radar. Leveraging spatio-temporal graph neural networks (ST-GNNs), it accurately estimated the hidden spatial normals of adjacent multi-frame point clouds, eliminating the need for position sensors. By transforming radar point cloud data from each frame into a unified observation coordinate system, the method facilitated multi-frame fusion of 4D point clouds and ensured precise scene alignment. Experimental results demonstrated that the proposed approach not only accurately assessed the spatial attitude of 4D point clouds but also effectively corrected and fused the coordinates of each point cloud frame. This enabled precise coordinate alignment during motion and vibration. Furthermore, the algorithm significantly enhanced point cloud imaging density, improved image accuracy and readability, and was capable of imaging both static and dynamic targets. It provided robust support for the application of millimeter-wave radar in outdoor motion scenes.