To address the shortcomings of wireless data transmission in aquaculture water quality monitoring, we have developed an intelligent water quality detection device based on the STM32 microcontroller system as the proce...
To address the shortcomings of wireless data transmission in aquaculture water quality monitoring, we have developed an intelligent water quality detection device based on the STM32 microcontroller system as the processing unit and TurMass transmission technology. this device can detect water quality in real-time and transmit the data to the user end for real-time monitoring so that users can monitor water quality in real-time. through rigorous and systematic experimental measurements, the accuracy of the detected data by the device was found to be extremely high, indicating that it can be used in practical applications.
Steganalysis is the opposite art to steganography, whose goal is to detect whether or not the seemly innocent objects like image hiding message. Withthe explosive growth of the Internet information. Making real time ...
详细信息
To meet the flexible and integrated requirements of future optical networks, we propose the optical filter with variable waveform shape using on-chip photonic reservoir computing neural network. the on-chip photonic r...
To meet the flexible and integrated requirements of future optical networks, we propose the optical filter with variable waveform shape using on-chip photonic reservoir computing neural network. the on-chip photonic reservoir computing neural network implements a hardware network directly to mitigate the latency and power-consumption. Combined with linear regression algorithm, infinite (IIR) and finite (FIR) impulse response optical filters are realized. Compared withthe traditional filters, this filter obtains more flexible and higher integration. In future, this filter can be widely used in integrated microwave photonics and wavelength division multiplexing systems.
Aiming at the problem that the detection of direct sequence spread spectrum (DSSS) signal under the jamming signals with different bandwidths, a detection method for DSSS signal based on decision tree is proposed in t...
Aiming at the problem that the detection of direct sequence spread spectrum (DSSS) signal under the jamming signals with different bandwidths, a detection method for DSSS signal based on decision tree is proposed in this paper. First, features are extracted as detection statistics based on cepstrum and time-domain cross-correlation, respectively. And then the decision tree is used to make the detection decision by comparing the detection statistics withthe predefined thresholds. After that, the optimal thresholds are determined through a global search algorithm, and the optimal thresholds are used for DSSS signal detection. the simulation results demonstrate that when the interference to signal ratio (ISR) is -1dB, the probability of detection reaches 95%.
the P300-based brain-machine interface can achieve efficient communication between human brain and computer, and help physically disabled people achieve autonomous control of external devices. In this paper, we propos...
the P300-based brain-machine interface can achieve efficient communication between human brain and computer, and help physically disabled people achieve autonomous control of external devices. In this paper, we propose a convolutional neural network-based signal detection method, which includes signal pre-processing, construction and training of convolutional neural network, and testing and performance evaluation of the model. this paper conducts experimental validation on dataset II of the 3rd BCI competition and compares it with some traditional machine learning methods and deep learning methods. the test accuracy of the experimental results reaches 0.85 and the AUC reaches 0.93, which indicates that the method has significantly improved in classification accuracy and model performance, and has high practicality and application prospects.
Criminal digital image processing technology is becoming increasingly important in public security operations, especially in the identification of characters in surveillance videos and the analysis of behavior and act...
Criminal digital image processing technology is becoming increasingly important in public security operations, especially in the identification of characters in surveillance videos and the analysis of behavior and actions in on-site videos, where blurry portraits in videos are particularly important. the paper mainly focuses on how to restore facial features of blurred human figures in video surveillance. Two methods, inverse filtering and Wiener filtering, are used to process blurred human figures, ultimately achieving the expected results. this provides assistance in identifying the same case as a major and difficult case.
作者：
Tan, DiFang, XiaohanFan, YuanAnhui University
Key Laboratory of Intelligent Computing and Signal Processing of Ministry of Education School of Electrical Engineering and Automation Hefei230601 China
Community microgrid is a kind of new power system that acts as an intermediary between community consumers and the grid, playing an important role in realising distributed autonomy and improving the power supply level...
详细信息
Ultra Wide Band radar life detection technology has been widely used in life detection in various scenes because of its high life detection accuracy, but it is difficult to extract vital signs in the process of echo s...
Ultra Wide Band radar life detection technology has been widely used in life detection in various scenes because of its high life detection accuracy, but it is difficult to extract vital signs in the process of echo signalprocessing because of scene ***, a research on the processing method of UWB radar echo signal is *** mean band-pass filter is used to remove and enhance the background noise of vital signs in UWB echo signal, and the vital signs signal is located by combining energy moment calculation. In order to improve the accuracy of vital sign signal extraction, and according to the linear characteristics of vital sign signals, an adaptive EWT algorithm is proposed by combining Pearson correlation coefficient with high computational efficiency and EWT algorithm with strict mathematical basis and no modal aliasing and endpoint effect problem, then the located vital sign signals are further filtered and extracted by this ***, the effectiveness of the proposed method is verified by experiments.
Traditional image emotion recognition focuses only on the emotion information embedded in the subject or part of the image, while ignoring the global emotion information. In this paper, we propose a new network based ...
Traditional image emotion recognition focuses only on the emotion information embedded in the subject or part of the image, while ignoring the global emotion information. In this paper, we propose a new network based on Vision Transformer: a global perception network, containing a main module and a background module. the main module enhances the weighting of objects by showing the interdependencies between modelled channels. the background module further searches for emotional information in the background. In addition, we employ multi-task learning and Adaptive Gradient Clipping to improve the network. Extensive experiments show that our proposed method outperforms the state-of-the-art methods for image sentiment analysis.
A recognition method based on feature processing and analysis of arc sound signals is proposed to address the difficulty of online detection of undercut defects in welding. Firstly, to eliminate redundant information ...
A recognition method based on feature processing and analysis of arc sound signals is proposed to address the difficulty of online detection of undercut defects in welding. Firstly, to eliminate redundant information in the sound signal, the signal is decomposed and reconstructed using the Haar wavelet basis function for wavelet packet decomposition. the relative wavelet energy in each frequency band is calculated. the maximum inter-class standard deviation is introduced to determine the sensitive frequency band for defect signals. Secondly, the reconstructed signal's time-domain and frequency-domain statistical features are extracted to construct the feature vector. Fisher criterion is used as the feature evaluation standard to determine the sensitive features. Finally, SVM (Radial base support vector machine) and five-fold cross-validation are used to establish a prediction model. the results show that the model can effectively predict and classify welding defects, with a recognition accuracy of 96.3%.
暂无评论