Lithium-ion batteries are widely used as rechargeable energy and power storage system in smart devices electric vehicles because of their high specific energy, densities, etc. The state charge (SOC) serves a vital feature that is monitored by the battery management to optimize performance, safety, lifespan lithium-ion batteries. In this paper, strong tracking adaptive fading-extended Kalman filter (STAF-EKF) based on second-order resistor–capacitor equivalent circuit model (2RC-ECM) proposed...

In order to alleviate the situation that small objects are prone missed detection and false in natural scenes, this paper proposed a object algorithm for adaptive feature fusion, referred as MMF-YOLO. First, aiming at problem pixels easy lose, multi-branch cross-scale fusion module with factor was proposed, where each path has an factor, which can allow network independently adjust importance of features according learned weights. Then, similar background information overlap complex M-CBAM...

An image encryption and decryption algorithm based on Arnod transform fractional chaos is proposed in this work for solving the problem that encrypted easily cracked content of decrypted distorted. To begin with, Arnold used to encrypt, so spatial confidence original has been comprehensively disturbed. Secondly, XOR involving order chaotic sequence encrypt. The key dynamically generated ensure randomness difference generation. When required, first performed using XOR. Then, second carried...

In order to overcome the problems of small proportion object, variable shooting angle, and height in UAV aerial image, an anchor-free object detection algorithm based on YOLOX network is proposed this paper. Firstly, aiming at problem size, a Tri-Head module change size head obtain more original information objects. Secondly, accurately detect dense continuous objects complex scenes, CBAM-CSP incorporating CBAM attention mechanism increase weights for detecting regions interest images. Then,...

Accurate expression interpretation occupies a huge proportion of human-to-human communication. The control expressions can facilitate more convenient communication between people. Expression recognition technology has also been transformed from relatively mature laboratory-controlled research to natural scenes research. In this paper, we design multi-channel attention network based on channel weighting for analysis in scenes. mainly consists three parts: Multi-branch feature extraction...

Abstract The automation and intelligence of grid generation are still bottleneck problems in computational fluid dynamics. This problem can be solved by generating anisotropic quadrilateral advancing layer method isotropic triangular mesh front method. However, the needs to calculate direction, step size, a correction factor concave convex area, so level efficiency need improved. To solve above problems, hybrid based on BP-ANN is proposed. data generated traditional methods learned neural...

Facial expressions occupy a large proportion of the communication between people. It is technology in which computers try to understand human emotions by analyzing face information, and it has become hot topic field computer vision. However, there are few reliable expression recognition databases task. The not robust enough under uncontrolled conditions, takes long time train facial model using deep learning method. This paper uses ResNet18 as backbone network construct An Adaptive Mixed...

Automatic detection of image damaged regions is the key to automatic video inpainting. Vertical scratches are common damages in old film. In this paper, a vertical algorithm based on edge proposed. The proposed first uses operator which has largest response Sobel detect edges, and then canny edges further. Third, we lines through probabilistic Hough transform. Finally, obtain true locations morphology width constraints. Many experiments show that our method can line effectively...

With the increasing of complexity computer networks, research network management system has become more and important. Firstly, in order to make performance satisfy need fault-tolerant, learning ability organizational ability, multilevel control model is researched this paper by combining with multi-Agent technology Multilevel theory. Then, on basis utilizing distribution collaboration Multi-Agent implement monitoring, analysis forecast performance, fulfilled through method. Thirdly,...

The video enhancement is necessary and needed in computer vision many applications, especially low illumination environment. In this paper, a fast method based on simplified VEC (Virtual Exposure Camera) model proposed to enhance dim videos. mainly includes two parts. One part the filters that we used pre post processing reduce noise. other tone mapping adjust illumination. Experimental results show effective real time processing, satisfied for exposure video.

This paper first analysis advantages and disadvantages of image zooming algorithm based on wavelat transform bilinear interpolation. Then according to the similarity bandelet component it gives a improved interpolation modified by transform. Experiments show that can properly retain abundant high frequency information in original image. After wavelet reconstruction we obtain an with higher resolution, better visual effect, Signal Noise Ratio(SNR), more detail no apparent warp.

The advancing front method (AFM) is one of the widely used unstructured grid generation techniques. However, efficiency relatively low because only cell generated in procedure. In this work, a novel automatic isotropic triangle technique developed by introducing an artificial neural network (ANN) based double-front (ADFM) to improve mesh efficiency. First, variety different patterns are extracted from AFM and extended ADFM method. process each pattern discussed detail. Second, initial...

The globalization pandemic of COVID-19 has made wearing masks to become a norm in people's lives, and this preventive measure brings new challenges face recognition algorithms. To address problem, paper, multi-branch network is proposed simultaneously complete the task masked detection recognition. Firstly, improves Swin Transformer for extraction facial features. Secondly, organ attention mechanism, FOA, make model focus on organs that are not covered by masks. Then, order overcome problem...

Deep multi-view subspace clustering (DMVSC) based on self-expression has attracted increasing attention due to its outstanding performance and nonlinear application. However, most existing methods neglect that view-private meaningless information or noise may interfere with the learning of self-expression, which lead degeneration performance. In this paper, we propose a novel framework Contrastive Consistency Attentive Complementarity (CCAC) for DMVsSC. CCAC aligns all self-expressions...

针对复杂背景下的快速行人检测问题，从行人边缘信息的角度，该文提出了一种改进的中心对称统计变换(ICS_CENTRIST)特征，该特征只有32维，计算简单，描述能力强，可以很好地表达行人的边缘轮廓。行人检测时采用3级级联分类方法：第1级采用基于辅助积分图的线性支持向量机(SVM)，快速排除大部分非行人区域；第2级，第3级分别使用偏最小二乘法(PLS)选出区分能力最强的前12和21个块(block)，提取ICS_CENTRIST特征，采用直方图交叉核支持向量机(HIK-SVM)进行精确检测。实验结果表明，该文算法在复杂背景下可取得较好的检测效果，检测速度在447358大小的图像上达到平均50 ms，与基于CENTRIST特征的快速检测方法和梯度方向直方图(HOG)算法相比分别提高了50%和90%，满足实时性要求。

The precise evaluation of camera position and orientation is a momentous procedure most machine vision tasks, especially visual localization. Aiming at the shortcomings local features dealing with changing scenes problem realizing robust end‐to‐end network that worked from feature detection to matching, an invariant matching method for scene image pairs proposed, which integrates detection, descriptor constitution, matching. In point construction stage, joint training carried out based on...

To improve the efficiency and automation of traditional advancing front method (AFM) unstructured grid generation, a novel isotropic triangular generation technique is developed based on an artificial neural network (ANN). First, some existing high-quality grids are used as data sources, then automatic extraction training dataset proposed. Second, input into ANN to train by back-propagation (BP) algorithm, typical patterns identified through iterative learning. Third, after inputting initial...

Most of the research on salient object detection pursues performance but ignores efficiency, resulting in poor practicability. In this paper, we propose a lightweight SOD solution, named GIPANet, which better alleviates contradiction between model size and performance. Firstly, to solve problems shallow depth insufficient information extraction feature networks, Global Pooling Aggregation Module (GPAM) is constructed. Secondly, Feature Enhancement (FAEM) constructed fuse features more...