光伏电池边缘检测算法研究

为了满足晶硅光伏电池视觉印刷高精度边缘定位要求,本文提出了一种改进的亚像素边缘定位算法.该算法首先应用Sobel算子和线性插值得到边缘垂直方向上新的边缘图像;最后对插值的边缘图像使用改进的灰度矩算子,得到的亚像素边缘.同时本文进行改进亚像素检测算法与传统亚像素检测算法的精度及鲁棒性对比实验.结果表明:本文算法比传统亚像素定位算法有着更好的定位精度和鲁棒性,直线检测精度能达到1μm;在高噪声污染的图像有着较高的检测精度,其标准差为0.195μm.     

刀具轮廓亚像素精度阈值分割算法研究

提取亚像素精度边缘轮廓是完成刀具几何参数精确测量的重要环节。该文研究了一种简单有效的方法,该算法基于灰度阈值分割定位像素级边缘,并经双线性插值法细分完成亚像素边缘轮廓的提取。最后设计实验对比分析了该算法以及基于Canny和Sobel算子的亚像素边缘检测算法所提取轮廓的特点。结果表明在准确的前提下,对高对比度图像应用该算法后,能比后两者更快速地完成边缘轮廓提取。     

基于SUSAN和Hough变换的直线边缘亚像素定位方法

提出了一种基于SUSAN算法和Hough变换的直线边缘亚像素定位方法.在该方法中,给出了SUSAN算法模板选择的依据,同时定义了直线边缘响应函数并引入加权Hough变换.首先,利用直线边缘响应函数对直线边缘进行提取;然后对具有响应值的灰度点进行Hough变换并将该响应值作为权值记入参数空间累加器,得到粗定位;在粗定位的基础上对映射区进行局部细化,并对区域内点进行拟合,最终得到直线边缘精定位.实验证明:直线边缘定位精度可达0.3 pixels,同时为解析曲线亚像素定位提供了一种新的思路.     

一种基于色相角算法的图像质量评价模型

目前计算图像色差的典型方法不能准确反应真实的观察效果。介绍了基于色相角算法的图像评价方法,从色相角度出发,对图像中的各个像素进行不同程度的加权,从而控制每个像素对整幅图像的影响效果。通过实验与传统色差计算方法进行了对比,验证了该方法的有效性和普遍性。     

图像中CCD光斑中心的亚像素定位研究

对图像中的目标进行定位是基于图像的动态测量和精密测量中的首要任务之一.本文首先采用迭代阈值法对图像中CCD光斑区域进行粗定位,然后用双线性插值法对截取的光斑图像进行插值处理,从而实现CCD光斑中心的亚像素定位.实验表明,该方法的定位精度可达1/10像素左右,大大提高了CCD光斑中心亚像素定位的精确度和稳定性.     

基于尺度不变特征变换特征的显微图像在线拼接方法

为解决显微镜观测视野狭小的问题,提出了一种基于尺度不变特征变换(SIFT)特征的显微图像自动拼接方法.该方法充分考虑了显微图像采集过程以及显微图像自身的特点,通过一系列适用于显微图像拼接的算法,控制单幅图像拼接时间在1秒内,进而可以保证显微图像拼接随图像采集在线完成,同时,利用SIFT特征匹配准确度高的优点,可达到亚像素级的拼接精度.该方法拓展了显微镜的应用,尤其适用于生物和医学显微图像观测.     

基于亚像素定位的贴装元件精确取像技术

自动光学检测系统中采集到的贴装元件图像存在平移和旋转误差.针对该问题,提出了基于亚像素定位技术的贴装元件精确取像方法,该方法结合Canny边缘分割、基于空间矩的亚像素细分、最小二乘法拟合等技术,对基准点进行亚像素定位,并以此为基础进行贴装元件的精确取像,消除电路板的定位误差和制造误差,从而保证了自动光学检测系统的检测准确度.实验结果表明:本文算法对元件图像提取精确且稳定,满足高精度视觉检测要求.     

几何重心法亚像素提取算法研究

研究了一种基于显微图像的几何重心法亚像素提取算法,用于实现超低转速的在线测量。首先根据显微图像法将超低转速的转动测量转换为平动测量; 然后利用提出的几何重心法的亚像素提取算法,解决了目前常用的曲线拟合法和梯度法的亚像素提取时间长、抗图像干扰能力弱的问题。几何重心法的核心是在整像素提取的基础上,在3×3的像素范围内,分别计算参考像素灰度值与提取的整像素灰度之差的指数值,在获取的3点指数值上,利用几何重心法提取出亚像素值。通过光栅基准平台微位移抓拍图像的验证,提出的几何重心法比曲面拟合法、梯度法用时短,平均用时<0.2s,提取的亚像素值的标准差<1/1000像素,满足了超低转速在线测量的基本要求。     

基于高斯积分曲线拟合的亚像素边缘提取算法

针对传统边缘提取算法定位精度低、对噪声敏感等缺点,提出一种基于高斯积分曲线拟合的亚像素边缘提取算法。通过曲面插值求取像素级边缘法截线上各离散点的灰度值,再进行高斯积分曲线拟合,寻找高斯积分曲线的均值点坐标,实现亚像素边缘的精定位。用量块直线边缘进行实验,并与现有亚像素边缘提取算法比较,实验证明基于高斯积分曲线拟合的亚像素边缘提取算法定位精度较高,可以达到1 μm,且算法可靠性高、计算速度快,能够用于高精度测量。     

基于改进Hough变换的直线图形快速提取算法

为能够有效解决实时直线图形提取问题,提出了一种基于多约束Hough变换(HT)的直线提取算法.该算法首先分析了数字图像中直线边缘的三种结构特征,提出采用基元结构表示目标边缘点,并在约束条件下计算基元结构的基元倾角.在此基础上,结合传统的HT的思想对基元结构进行极角约束HT,以获得最终的直线参数.实验结果表明,对合成图像和自然图像,该算法比梯度HT的运算速度分别提高约190倍和22倍.     

基于门牌号识别的移动机器人全局自定位方法研究

提出了一种采用单目视觉和多超声传感器来识别门牌号的室内移动机器人全局自定位方法和导航策略。机器人通过搜索并识别走廊中各房间的门牌号码来实现自身的定位与导航。此外，给出了该导航系统的机器人控制体系结构。依据该体系结构，运用Hough变换，Randon变换和模板匹配等手段，给出了搜索策略及有关算法，主要包括门的定位与识别、图像的倾斜校正、图像分割以及门牌号的识别等。最后，利用我们自行设计和开发的CASIA-I移动机器人对所提出的导航策略和导航算法进行了实物实验验证。结果表明该导航策略是一种较为有效的方法。     

Automatic Vehicle License Plate Recognition Using Optimal Deep Learning Model

The latest advancements in highway research domain and increase inthe number of vehicles everyday led to wider exposure and attention towards the development of efficient Intelligent Transportation System (ITS). One of the popular research areas i.e., Vehicle License Plate Recognition (VLPR) aims at determining the characters that exist in the license plate of the vehicles. The VLPR process is a difficult one due to the differences in viewpoint, shapes, colors, patterns, and non-uniform illumination at the time of capturing images. The current study develops a robust Deep Learning (DL)-based VLPR model using Squirrel Search Algorithm (SSA)-based Convolutional Neural Network (CNN), called the SSA-CNN model. The presented technique has a total of four major processes namely preprocessing, License Plate (LP) localization and detection, character segmentation, and recognition. Hough Transform (HT) is applied as a feature extractor and SSA-CNN algorithm is applied for character recognition in LP. The SSA-CNN method effectively recognizes the characters that exist in the segmented image by optimal tuning of CNN parameters. The HT-SSA-CNN model was experimentally validated using the Stanford Car, FZU Car, and HumAIn 2019 Challenge datasets. The experimentation outcome verified that the presented method was better under several aspects. The projected HT-SSA-CNN model implied the best performance with optimal overall accuracy of 0.983%.     

基于Hough变换的三次方Bezier曲线检测算法研究

类似经典Hough变换中对直线(段)、圆(弧)、椭圆、抛物线等解析曲线的检测,论文研究了三次方Bezier曲线的检测算法,提出了离散Bezier曲线的特征建模方法和使用R函数的Hough变换曲线检测快速算法。该算法能够根据所给出的待检测目标点阵图像建立形状参数模型,然后检测该曲线在复杂图像中出现的位置、大小和方向。实验表明,该法能够有效地检测任意三次方Bezier曲线,且精确度优于目前广泛用于曲线检测的广义Hough变换。     

Three-Dimensional Distance-Error-Correction-Based Hop Localization Algorithm for IoT Devices

The Internet of Things (IoT) is envisioned as a network of various wireless sensor nodes communicating with each other to offer state-of-the-art solutions to real-time problems. These networks of wireless sensors monitor the physical environment and report the collected data to the base station, allowing for smarter decisions. Localization in wireless sensor networks is to localize a sensor node in a two-dimensional plane. However, in some application areas, such as various surveillances, underwater monitoring systems, and various environmental monitoring applications, wireless sensors are deployed in a three-dimensional plane. Recently, localization-based applications have emerged as one of the most promising services related to IoT. In this paper, we propose a novel distributed range-free algorithm for node localization in wireless sensor networks. The proposed three-dimensional hop localization algorithm is based on the distance error correction factor. In this algorithm, the error decreases with the localization process. The distance correction factor is used at various stages of the localization process, which ultimately mitigates the error. We simulated the proposed algorithm using MATLAB and verified the accuracy of the algorithm. The simulation results are compared with some of the well-known existing algorithms in the literature. The results show that the proposed three-dimensional error-correction-based algorithm performs better than existing algorithms.     

Methods to detect objects in photon-limited images

We investigate the problem of detecting and localizing a known signal in a photon-limited image, where Poisson noise is the dominant source of image degradation. For this purpose we developed and evaluated three new algorithms. The first two are based on the impulse restoration (IR) principle and the third is based on the generalized likelihood ratio test (GLRT). In the IR approach, the problem is formulated as one of restoring a delta function at the location of the desired object. In the GLRT approach, which is a well-known variation on the optimal likelihood ratio test, the problem is formulated as a hypothesis testing problem, in which the unknown background intensity of the image and the intensity scale of the object are obtained by maximum-likelihood estimation. We used Monte Carlo simulations and localization receiver operating characteristic (LROC) curves to evaluate the proposed algorithms quantitatively. LROC curves demonstrate the ability of an algorithm to detect and locate objects in a scene correctly. Our simulations demonstrate that the GLRT approach is superior to all other tested algorithms.     

Position Vectors Based Efficient Indoor Positioning System

With the advent and advancements in the wireless technologies, Wi-Fi fingerprinting-based Indoor Positioning System (IPS) has become one of the most promising solutions for localization in indoor environments. Unlike the outdoor environment, the lack of line-of-sight propagation in an indoor environment keeps the interest of the researchers to develop efficient and precise positioning systems that can later be incorporated in numerous applications involving Internet of Things (IoTs) and green computing. In this paper, we have proposed a technique that combines the capabilities of multiple algorithms to overcome the complexities experienced indoors. Initially, in the database development phase, Motley Kennan propagation model is used with Hough transformation to classify, detect, and assign different attenuation factors related to the types of walls. Furthermore, important parameters for system accuracy, such as, placement and geometry of Access Points (APs) in the coverage area are also considered. New algorithm for deployment of an additional AP to an already existing infrastructure is proposed by using Genetic Algorithm (GA) coupled with Enhanced Dilution of Precision (EDOP). Moreover, classification algorithm based on k-Nearest Neighbors (k-NN) is used to find the position of a stationary or mobile user inside the given coverage area. For k-NN to provide low localization error and reduced space dimensionality, three APs are required to be selected optimally. In this paper, we have suggested an idea to select APs based on Position Vectors (PV) as an input to the localization algorithm. Deducing from our comprehensive investigations, it is revealed that the accuracy of indoor positioning system using the proposed technique unblemished the existing solutions with significant improvements.     

小包烟包装质量机器视觉检测关键技术研究

针对烟草企业对小包烟包装质量精密检测的需要,提出基于机器视觉的检测系列关键技术,包括了小包烟图像采集装置、基于快速Hough变换的小包边缘快速提取算法、基于灰度共生矩(GLCM)参数快速在线判别算法、基于支持向量机(SVMs)的包装缺陷识别等.该关键技术不仅适应在线快速包装质量检测,而且能准确判断缺陷类型.实验证明该关键技术具有实用可靠等优点,可应用到小包烟生产现场.     

Localization Based Evolutionary Routing (LOBER) for Efficient Aggregation in Wireless Multimedia Sensor Networks

Efficient aggregation in wireless sensor nodes helps reduce network traffic and reduce energy consumption. The objective of this work Localization Based Evolutionary Routing (LOBER) is to achieve global optimization for aggregation and WMSN lifetime. Improved localization is achieved by a novel Centroid Based Octant Localization (CBOL) technique considering an arbitrary hexagonal region. Geometric principles of hexagon are used to locate the unknown nodes in the centroid positions of partitioned regions. Flower pollination algorithm, a meta heuristic evolutionary algorithm that is extensively applied in solving real life, complex and nonlinear optimization problems in engineering and industry is modified as Enhanced Flower Pollination Algorithm (EFPA) to fit into WMSN and enhance routing mechanism and ensure efficiency in data aggregation. The system is simulated using MATLAB and found to have a considerable improvement in the optimization process.     

基于能量守恒全零DCT系数块的预先判别算法

针对全零DCT系数块,为避免对其进行变换和量化操作以减少编码器的运算量,从DCT公式、量化和能量守恒出发,在分析现有判别法对全零DCT系数块的误判与漏判的基础上,提出了一种预先判断全零DCT系数块的新方法,可应用于基于块的视频编码器中.同时,结合H.264整数变换和量化的特点对该算法进行了改进和仿真实验.实验结果表明,新提出的算法在保持PSNR不变的情况下,较现有典型方案有高出达32%的检出率,因而可更大幅度地降低DCT变换量化模块的计算复杂度.