Search

The flow pattern is one of the most significant parameters in modeling the oil–water two-phase system. How to extract efficient and objective features to precisely identify the oil–water two-phase flow patterns is still a significant issue. Inspired by the deep learning hierarchically feature extraction way, we, in this paper, employ convolutional neural networks (CNNs) to identify oil–water two-phase flow patterns. First, we carry out oil–water two-phase flow experiment and collect different oil–water flow pattern images. Then, we propose an image segment algorithm based on the minimum gray level to obtain the interest region that reflects the flow pattern characteristics. Finally, we employ three frequently used CNNs, LeNet-5, AlexNet, and VGG-16 net, to extract the image features and identify typical oil–water two-phase flow patterns. The results show that networks with more deep structures preserve relatively high flow pattern recognition accuracy. This paper provides a novel application of the deep learning method for the oil–water two-phase flow identification.

In order to protect the copyright of the color image, a robust color image watermarking method based on Tensor-Singular Value Decomposition (T-SVD) is proposed. In order to preserve strong correlations of the RGB three channels of the color image, the color image is considered as the third-order tensor, which is transformed by using T-SVD to obtain two orthogonal tensors and one diagonal tensor. Since the diagonal tensor includes the main energies of the color image, it can be exploited to embed watermark for robustness. Firstly, the three channels of the color image are decomposed by using discrete wavelet transform (DWT) into four subbands, respectively, and the LL subband of each channel is used to construct an approximation image. Secondly, the approximation image is divided into non-overlapping blocks with the size of 4 × 4, and the TSVD is applied in each block to compute the first, the second and the third diagonal matrices. Finally the grayscale watermark is decomposed by using Singular Value Decomposition (SVD). The principal components of the gray-scale watermark are embedded by modifying the first diagonal matrix of each block, since the first diagonal matrix is more robust than other two diagonal matrices. Experiment results show that the proposed method can resist a variety of image attacks, and is more robust than the existing watermarking methods.

Sparse Multinomial Logistic Regression (SMLR) is widely used in the field of image classification, multi-class object recognition, and so on, because it has the function of embedding feature selection during classification. However, it cannot meet the time and memory requirements for processing large-scale data. We have reinvestigated the classification accuracy and running efficiency of the algorithm for solving SMLR problems using the Alternating Direction Method of Multipliers (ADMM), which is called fast SMLR (FSMLR) algorithm in this paper. By reformulating the optimization problem of FSMLR, we transform the serial convex optimization problem to the distributed convex optimization problem, i.e., global consensus problem and sharing problem. Based on the distributed optimization problem, we propose two distribute parallel SMLR algorithms, sample partitioning-based distributed SMLR (SP-SMLR), and feature partitioning-based distributed SMLR (FP-SMLR), for a large-scale sample and large-scale feature datasets in big data scenario, respectively. The experimental results show that the FSMLR algorithm has higher accuracy than the original SMLR algorithm. The big data experiments show that our distributed parallel SMLR algorithms can scale for massive samples and large-scale features, with high precision. In a word, our proposed serial and distribute SMLR algorithms outperform the state-of-the-art algorithms.

The spatial-temporal baseline threshold method is commonly used to select interferograms for the Small BAseline Subset interferometric synthetic aperture radar (SBAS-InSAR) technique. However, this selection strategy is rather empirical and prone to including highly contaminated interferograms or excluding those with high quality. To overcome these limitations, this study first derives the relationship between the measurement accuracy of unknown parameters and the number of selected interferograms with their corresponding qualities. Subsequently, an adaptive interferogram selection method is proposed on the basis of Graph Theory (GT) and the turbulence atmospheric effects of interferogram. This proposed method first identifies and deletes the SAR image that is severely polluted by atmospheric phase. Second, high-quality interferograms are selected for SBAS-InSAR based on their corresponding turbulence atmospheric variance. Compared with the traditional selection method, this approach can significantly reduce the effect of turbulence atmosphere on SBAS-InSAR. A set of simulated experiments and real Sentinel-1A data in Hawaii, United States, validate the good performance of the proposed method.