Deep Hash with Improved Dual Attention for Image Retrieval

Recently, deep learning to hash has extensively been applied to image retrieval, due to its low storage cost and fast query speed. However, there is a defect of insufficiency and imbalance when existing hashing methods utilize the convolutional neural network (CNN) to extract image semantic features and the extracted features do not include contextual information and lack relevance among features. Furthermore, the process of the relaxation hash code can lead to an inevitable quantization error. In order to solve these problems, this paper proposes deep hash with improved dual attention for image retrieval (DHIDA), which chiefly has the following contents: (1) this paper introduces the improved dual attention mechanism (IDA) based on the ResNet18 pre-trained module to extract the feature information of the image, which consists of the position attention module and the channel attention module; (2) when calculating the spatial attention matrix and channel attention matrix, the average value and maximum value of the column of the feature map matrix are integrated in order to promote the feature representation ability and fully leverage the features of each position; and (3) to reduce quantization error, this study designs a new piecewise function to directly guide the discrete binary code. Experiments on CIFAR-10, NUS-WIDE and ImageNet-100 show that the DHIDA algorithm achieves better performance.

Ultrasonic unilateral double-position excitation lamb wave defect detection and quantification method for ground electrode of transmission tower

The power transmission tower’s ground electrode defect will affect its normal current dispersion function and threaten the power system’s safe and stable operation and even personal safety. Aiming at the problem that the buried grounding grid is difficult to be detected, this paper proposes a method for identifying the ground electrode defects of transmission towers based on single-side multi-point excited ultrasonic guided waves. The geometric model, ultrasonic excitation model, and physical model are established, and the feasibility of ultrasonic guided wave detection is verified through the simulation and experiment. In actual inspection, it is equally important to determine the specific location of the defect. Therefore, a multi-point excitation method is proposed to determine the defect’s actual position by combining the ultrasonic guided wave signals at different excitation positions. Besides, the precise quantification of flat steel grounding electrode defects is achieved through the feature extraction-neural network method. Field test results show that, compared with the commercial double-sided excitation transducer, the single-sided excitation transducer proposed in this paper has a lower defect quantization error in defect quantification. The average quantization error is reduced by approximately 76%.

Electrocardiogram transmission over OFDM system

Remote patient monitoring is an important area of research. Electrocardiogram (ECG) is a vital health parameter which can be remotely transmitted to monitor the patient’s health. In this field, there are many research directions which include ECG security, patient data hiding in ECG, ECG classification, ECG transmission and reception. In this paper, an effective methodology for ECG transmission over OFDM wireless communication system has been presented. Issues related to transmission of ECG as image have also been discussed. Before ECG is transmitted over OFDM analog ECG signal is to be converted into series of bits for which sampling and quantization has to be performed. The quantization error arises due to number of chosen samples and quantization levels. In this work, it is shown that the quantization error can be brought down to zero using symbolic aggregate approximation (SAX) for data representation. Performance of the ECG transmission using different methodologies has been compared using peak signal-to-noise ratio (PSNR) Structural Similarity Index Metric (SSIM), Bit Error Rate (BER) and Pixel Error Rate (PER).