Image Augmentation-Based Food Recognition with Convolutional Neural Networks

Image retrieval for food ingredients is important work, tremendously tiring, uninteresting, and expensive. Computer vision systems have extraordinary advancements in image retrieval with CNNs skills. But it is not feasible for small-size food datasets using convolutional neural networks directly. In this study, a novel image retrieval approach is presented for small and medium-scale food datasets, which both augments images utilizing image transformation techniques to enlarge the size of datasets, and promotes the average accuracy of food recognition with state-of-the-art deep learning technologies. First, typical image transformation techniques are used to augment food images. Then transfer learning technology based on deep learning is applied to extract image features. Finally, a food recognition algorithm is leveraged on extracted deep-feature vectors. The presented image-retrieval architecture is analyzed based on a small-scale food dataset which is composed of forty-one categories of food ingredients and one hundred pictures for each category. Extensive experimental results demonstrate the advantages of image-augmentation architecture for small and medium datasets using deep learning. The novel approach combines image augmentation, ResNet feature vectors, and SMO classification, and shows its superiority for food detection of small/medium-scale datasets with comprehensive experiments.     

Multiple Kernel Clustering Based on Self-Weighted Local Kernel Alignment

Multiple kernel clustering based on local kernel alignment has achieved outstanding clustering performance by applying local kernel alignment on each sample. However, we observe that most of existing works usually assume that each local kernel alignment has the equal contribution to clustering performance, while local kernel alignment on different sample actually has different contribution to clustering performance. Therefore this assumption could have a negative effective on clustering performance. To solve this issue, we design a multiple kernel clustering algorithm based on self-weighted local kernel alignment, which can learn a proper weight to clustering performance for each local kernel alignment. Specifically, we introduce a new optimization variable- weight-to denote the contribution of each local kernel alignment to clustering performance, and then, weight, kernel combination coefficients and cluster membership are alternately optimized under kernel alignment frame. In addition, we develop a three-step alternate iterative optimization algorithm to address the resultant optimization problem. Broad experiments on five benchmark data sets have been put into effect to evaluate the clustering performance of the proposed algorithm. The experimental results distinctly demonstrate that the proposed algorithm outperforms the typical multiple kernel clustering algorithms, which illustrates the effectiveness of the proposed algorithm.     

A Secure Rotation Invariant LBP Feature Computation in Cloud Environment

In the era of big data, outsourcing massive data to a remote cloud server is a promising approach. Outsourcing storage and computation services can reduce storage costs and computational burdens. However, public cloud storage brings about new privacy and security concerns since the cloud servers can be shared by multiple users. Privacy-preserving feature extraction techniques are an effective solution to this issue. Because the Rotation Invariant Local Binary Pattern (RILBP) has been widely used in various image processing fields, we propose a new privacy-preserving outsourcing computation of RILBP over encrypted images in this paper (called PPRILBP). To protect image content, original images are encrypted using block scrambling, pixel circular shift, and pixel diffusion when uploaded to the cloud server. It is proved that RILBP features remain unchanged before and after encryption. Moreover, the server can directly extract RILBP features from encrypted images. Analyses and experiments confirm that the proposed scheme is secure and effective, and outperforms previous secure LBP feature computing methods.     

智能算法在交流调速系统中的研究

提出一种基于遗传算法优化的异步电机参考模型模糊自适应矢量控制调速系统,其异步电机模糊自适应速度辨识机构可根据参考模型输出与被控可调模型输出之差及其变化率,产生一个模糊自适应信号,控制被控可调模型的输出趋于参考模型的输出.仿真和实验表明,该系统具有良好的动态特性,在稳态精度上也优于普通的模型参考自适应系统,为无速度传感器...     

Birdsong classification based on multi-feature fusion

Multimedia Tools and Applications - The classification of birdsong has very important signification to monitor the bird population in the habitats. Aiming at the birdsong dataset with complex and...     

A modeling and control approach to magnetic levitation system based on state-dependent ARX model

Magnetic levitation (Maglev) systems are usually strongly nonlinear, open-loop unstable and fast responding. In order to control the position of the steel ball in a Maglev system, a data-driven modeling approach and control strategy is presented in this paper. A state-dependent AutoRegressive with eXogenous input (SD-ARX) model is built to represent the dynamic behavior between the current of electromagnetic coil and the position of the ball. State-dependent functional coefficients of the SD-ARX model are approximated by Gaussian radial basis function (RBF) neural networks. The model parameters are identified offline by applying the structured nonlinear parameter optimization method (SNPOM). Based on the model, a predictive controller is designed to stabilize the magnetic levitation ball to a given position or to make it track a desired trajectory. The real-time control results of the proposed approach and the comparisons with other two approaches are given, which demonstrate that the modeling and control method presented in this paper are very effective and superior in controlling the fast-responding, strongly nonlinear and open-loop unstable system. This paper gives the real experimental evidence that the RBF-ARX model is capable of not only globally, but also locally capturing and quantifying a nonlinear and fast-response system's behavior, and the model-based predictive control strategy is able to work quite well in a wide working-range of the nonlinear system.     

基于机器视觉的罐盖质量检测系统设计

文中根据饮料易拉罐罐盖制造生产线的工作环境和检测要求,研制了基于机器视觉的罐盖质量检测系统,实现了铝制罐盖瑕疵的自动检测和快速剔除.该检测系统由下盖装置、盖传送装置、光源与图像采集系统、视觉处理及控制系统、次品剔除装置等组成,铝制罐盖经下盖装置连续不断的进入盖传输区域,盖传输装置通过真空将罐盖吸附在传送带上,当罐盖通过成像系统时,光纤传感器触发工业相机和光源,获得高速罐盖图像,图像检测系统分析罐盖多个检测区域,电气控制系统根据图像检测结果分拣罐盖.通过实验测试证明:该视觉系统实时性好,可靠性高,有效地提高了罐盖检测生产线的工作效率.     

数据流上一种单遍扫描频繁模式树结构

针对频繁模式增长算法无法适应数据流的无限性和流动性的特点,提出一种新颖的FP-tree的变形结构--FPS-tree,只需单遍扫描便能获取当前窗口的全部数据库信息。为了在滑动窗口时有效地删除过期窗格和插入新窗格,提出一个新颖的概念--“尾结点”,FPS-tree中每条路径上的窗格信息只保持在尾结点里。实验结果表明FPS-tree的压缩性能要优于其他单遍扫描的前缀树结构。     

Predicate control: synchronization in distributed computations with look-ahead

The predicate control problem involves synchronizing a distributed computation to maintain a given global predicate. In contrast with many popular distributed synchronization problems such as mutual exclusion, readers writers, and dining philosophers, predicate control assumes a look-ahead, so that the computation is an off-line rather than an on-line input. Predicate control is targeted towards applications such as rollback recovery, debugging, and optimistic computing, in which such computation look-ahead is natural.We define predicate control formally and show that, in its full generality, the problem is NP-complete. We find efficient solutions for some important classes of predicates including “disjunctive predicates”, “mutual exclusion predicates”, and “readers writers predicates”. For each class of predicates, we determine the necessary and sufficient conditions for solving predicate control and describe an efficient algorithm for determining a synchronization strategy. In the case of “independent mutual exclusion predicates”, we determine that predicate control is NP-complete and describe an efficient algorithm that finds a solution under certain constraints.