Automatically assigning relevant text keywords to images is an important problem. Many algorithms have been proposed in the past decade and achieved good performance. Efforts focused upon model representations of keywords, but properties features not well investigated. In most cases, a group preselected, yet feature are used select features. this paper, we introduce regularization based selection algorithm leverage both sparsity clustering features, incorporate it into image annotation task....

Retrieving an occluded pedestrian remains a challenging problem in person re-identification (re-id). Most existing methods utilize external detectors to disentangle the visible body parts. However, these are unstable due domain bias and consume numerous computing resources. In this paper, we propose novel lightweight Part-based Representation Enhancement (PRE) network for re-id that takes full advantages of local correlations aggregate distinctive information features without relying on...

Our research addresses the critical intersection of communication and power systems in era advanced information technologies. We highlight strategic importance base station placement, as its optimization is vital for minimizing operational disruptions energy systems. study introduces a communications coordination planning (CPCP) model that encompasses both distributed resources stations to improve quality service. This facilitates optimal resource distribution, ensuring reliability over 96%...

As manufacturing processes continue to evolve, distributed and energy-efficient flexible systems have become the central paradigm of intelligent manufacturing. To effectively meet demands intricate production scheduling requirements, it is crucial take into account practical factors, such as two-stage assembly finite transportation resources. However, this imposes greater on efficiency algorithms. In paper, a knowledge-based bi-hierarchical optimization algorithm (KBOA) specially designed...

Rolling bearings are widely used in rotating machinery, such as aero-engine spindles, flying machines, wind turbines, etc. Bearing condition monitoring is of practical importance. The acoustic emission (AE) signal has impact and rapid attenuation characteristics. Most existing research on fault diagnosis not focused According to this characteristic, a time-frequency coherent energy change rate (TFC-TFECR) method proposed identify the AE signals bearing faults. This paper investigates effect...

Change Detection (CD) is an important research topic in the remote sensing field, and it has a wide range of applications, including resource monitoring, disaster assessment, urban planning, etc. Recently, Deep Learning (DL) shown its advantages CD. However, most existing DL-based methods can not capture complementary information between bi-temporal difference features. This paper proposes Edge-Guided Parallel Network (EGPNet) to solve this problem. First, our EGPNet extracts features...

Remote monitoring of bearing conditions, using wireless sensor network (WSN), is a developing trend in the industrial field. In complicated environments, WSN face three main constraints: low energy, less memory, and operational capability. Conventional data-compression methods, which concentrate on data compression only, cannot overcome these limitations. Aiming at problems, this paper proposed compressed acquisition reconstruction scheme based Compressed Sensing (CS) novel signal-processing...

We consider the problem of subspace clustering using SSC (Sparse Subspace Clustering) approach, which has several desirable theoretical properties and been shown to be effective in various computer vision applications.We develop a large scale distributed framework for computation via an alternating direction method multiplier (ADMM) algorithm. The proposed solves column blocks only involves parallel multivariate Lasso regression subproblems sample-wise operations. This appealing property...

This paper presents an algorithm to classify pixels in uterine cervix images into two classes, namely normal and abnormal tissues, simultaneously select relevant features, using group sparsity. Because of the large variations image appearance due changes illumination, specular reflections other visual noise, classes have a strong overlap feature space, whether features are obtained from color or texture information. Using more makes separable increases segmentation's quality, but also its...

As a spontaneous facial expression, micro-expression can reveal the psychological responses of human beings. Thus, recognition be widely studied and applied for its potentiality in clinical diagnosis, research, security. However, is formidable challenge due to short-lived time frame low-intensity actions. In this paper, sparse spatiotemporal descriptor developed by using Enhanced Local Cube Binary Pattern (Enhanced LCBP). The proposed LCBP composed three complementary binary features...

The control theory of self-adaptable fuzzy-PID is expounded in this paper, the dynamic characteristic and system structure boiler drum water level introduced also. method applied automatic level, under situation interaction no interaction, simulation research done to PID fuzzy-PID, result shows that application superior apparently.

Image saliency detection is one of the most active research topics in field computer vision. It focus on how to detect significant objects under complex background, and reduce computational cost for getting high-resolution, clear boundary, overall uniform objects. First, this paper state-of-the-art image analyzed presented detail. Then related methods are classified into two types space domain based, frequency prototypical tested public databases detection. Finally, development tendency predicted.

In this paper, we propose an automated, data-driven and unobtrusive framework to analyze interactional synchrony. We use information determine whether interpersonal synchrony can be indicator of deceit. Our includes a robust facial tracking module, effective expression recognition method, feature extraction selection methods. These features are used learn classification models for the deception recognition. To evaluate our proposed framework, have conducted extensive experiments on database...

For stereo vision applications, projective geometry has proved to be a useful tool for solving the rectification problem without camera calibration. However, criterion of minimisation must chosen properly in order avoid unduly geometric distortion. In this paper, an improved algorithm minimise distortion by combining newly developed transform with shearing is proposed. The emphasis on low makes method not only appropriate 3-D reconstruction but also stereoscopic viewing applications. On...

Sparse subspace clustering (SSC) is a classical method to cluster data with specific structure for each group. It has many desirable theoretical properties and been shown be effective in various applications. However, under the condition of large-scale dataset, learning sparse sample affinity graph computationally expensive. To tackle computation time cost challenge, we develop memory-efficient parallel framework computing SSC via an alternating direction multiplier (ADMM) algorithm. The...

In machine vision and the vehicle recognition system, removal of moving shadows is a significant topic. this paper, we propose novel method to detect in traffic video sequences. Firstly, set regions are segmented from sequence using background subtraction technique. Secondly, fast normalized cross-correlation (FNCC) adopted grayscale By utilizing three sum-table schemes, FNCC algorithm dramatically reduces computational complexity compared traditional cross correlation (NCC) algorithm. And...

The wind turbine failure-prone transmission system (shaft, gearbox, high-speed axis and generator) were regarded as the research object in this article. A method based on self-organizing fuzzy clustering Elman network faults diagnosis was proposed, which preprocessing data samples with criterion collected by vibration sensors turbine, analysis neural network. experimental results showed that diagnostic accuracy of has been improved comparing traditional method.

Due to massive date be monitored for Metro shield machine, in order solve the problems of knowledge acquisition bottlenecks and complexity structure network long traing time which based on expert system neural fault diagnosis methods. This article will introduces rough set theory subway machine diagnosis, Propose a method combine with diagnosis. Use strong advantage sets data classification, Remove redundancy information not effective decision-making. Then uses reduced as sample. Application...