基于视觉的车位线识别算法

提出了一种自动泊车系统中采用视觉方法通过识别车位线来确定泊车位的算法。采用金字塔分层搜索策略,首先,在灰度直方图上应用K均值聚类法对图像进行二值化,提取车位线骨架,采用Hough变换检测骨架,并利用基于密度的无参数聚类方法对骨架线聚类,在金字塔高层图像上确定车位角点候选点;然后,在金字塔最底层图像上选择感兴趣区域,采用改进的基于距离变换的骨架提取算法提取骨架,使用遗传算法对车位角点骨架进行精确匹配,根据实际车位角点的分布特征确定目标车位;最后,在室外不同环境下采集多张车位图片进行算法的有效性和快速性验证实验。实验结果表明,采用基于视觉的车位线识别算法进行车位检测能较大地提高检测的效率和识别正确率。     

基于聚类-改进灰狼算法的设施番茄分割识别方法

设施环境中的果实图像识别是作物生长监控和采摘智能化的关键技术之一。以设施番茄(品种:苏粉14)为研究对象,针对其获取图像果实粘连导致机器作业识别不准确,且识别效果受光线影响等问题,提出了一种改进灰狼算法优化聚类中心的番茄图像自动分割方法,并通过形态学和多尺度Hough变换算法对目标进行提取,完成图像中粘连果实的分割与识别。首先通过添加扰动因子对灰狼算法进行改进,调整搜索半径,增强全局寻优能力,避免算法陷入局部最优,然后利用改进灰狼算法对K均值聚类算法的聚类中心进行优化,模拟灰狼狩猎机制,寻找一组最佳聚类中心,克服原始聚类算法对初始中心点的过度依赖,提高其聚类效果,为识别设施番茄果实提供了良好的分割基础。对分割后的图像进行几何形态学处理,然后通过多尺度Hough圆检测,对果实进行识别和定位。随机选取200幅番茄图像进行实验,实验结果表明,相对于粒子群优化算法和遗传算法等优化算法,聚类-改进灰狼算法在收敛速度、聚类质量和稳定性方面都有显著提升,其中运行时间比K均值聚类算法缩短48.7%,峰值信噪比提高了45.2%,结构相似度提升了10.2%。聚类-改进灰狼算法果实平均识别率为89.2%,比通过K均值聚类算法进行分割的果实平均识别率提高了10.8%,识别率明显提高,实现了番茄果实的识别和定位。     

动态多目标优化的运动物体图像分割

对小区背景下运动物体图像进行分割时多使用单目标或多目标优化方法,这类方法不能有效适应目标的动态变化,因此本文提出一种动态多目标图像分割优化方法。该方法将时间及环境动态因素作为动态因子,利用K均值(KMeans)算法和和模糊C均值(FCM)聚类算法构造多目标函数;结合动态多目标粒子群算法(DMPSO),使用背景差分法定义环境变化规则,实现动态多目标的图像分割。根据DMPSO算法优化后的聚类结果,分别与K-Means和FCM聚类方法得到的结果进行了对比。结果表明,动态多目标优化的Pareto最优解集分布均匀,图像分割准确率可达到95%,对图像识别的准确率可达到90%,具有较高的识别能力,能满足确定背景下运动物体的准确识别。     

结合密度峰聚类的K均值图像分割算法

针对K均值聚类算法在图像分割应用中的不足,结合密度峰聚类算法对原有算法进行改进,得到了一种图像分割效果较好的改进K均值算法。K均值算法需要人工指定聚类中心数目,并且聚类中心的随机初始化对最终的图像分割结果有很大影响。针对以上缺点,对K均值算法进行改进,通过密度峰聚类算法自动确定了图像分割的聚类中心数目和较为准确的初始聚类中心。为了衡量色差在人眼中的感知情况,在算法中引入了NBS距离作为距离测度。实验结果表明,改进后的图像分割算法在分割图像时具有稳定的性能和较好的效果。     

基于MAR与FCM聚类的声呐图像分割

针对模糊C均值聚类算法中,聚类效果往往受到聚类中心数目和初始聚类中心的影响这一问题,提出一种基于多尺度自回归(MAR)模型与模糊C均值(FCM)聚类的声呐图像分割方法.引入MAR模型,建立层与层之间以及相邻层像素点间的数学关系,利用粗尺度图像的灰度-邻域均值二维直方图中的峰值个数来确定聚类中心数目,通过MAR得到的预测分割结果引导初始聚类中心的确定.实验结果表明,改进后的算法能准确、快速地确定聚类中心数目,并较好地解决初始聚类中心问题;与传统的FCM聚类方法相比,具有分割准确和收敛速度快的特点.     

七彩LED金丝球焊机图像识别系统设计

在对七彩LED引线键合工艺及芯片图像特征分析的基础上,通过芯片、焊点示教方案以及目标搜索、对中方案的规划,设计了一个七彩LED芯片、焊点自动识别及自动对中系统.将图像识别系统应用于七彩LED金丝球焊机上,实现了多芯片、多引线电子元器件的自动焊线作业,提高了生产效率,降低了焊线成本.     

实时在线桥梁模态参数自动识别

为系统实现桥梁实时在线模态参数自动识别,从数据分析和系统搭建两方面解决桥梁模态参数自动识别问题。首先,引入改进的模糊C均值聚类(fuzzy C-means,简称FCM)算法,通过设定最大聚类数目,得到不同聚类数目下的累积邻接矩阵,并引入图切分算法,自动确定最佳聚类数目,实现稳定图自动识别;其次,提出集数据采集、传输和分析一体化的模态参数自动识别系统框架,通过建立数据解析过程,保证实时在线模态识别的数据获取,并提出动态可变滑动数据窗,保证在线自动识别的实时性;最后,将提出的算法和系统框架运用于拱桥模型来验证所提方法和系统的可行性。结果表明:改进的模糊C均值聚类算法不需要任何人工干预,在默认参数下就可以自动获得最佳聚类数目,实现稳定图自动识别;提出的自动识别系统框架和动态可变滑动数据窗能够实现模态参数自动识别,并保证识别的实时性;提出的桥梁模态参数自动识别方法和自动识别系统能够用于桥梁结构实时在线模态参数的自动识别。     

改进的遗传算法在实时图像分割中的应用

为了自动确定图像分割的最佳阈值,提出了一种改进的自适应遗传算法,并利用该算法对二维Fisher准则图像分割评价函数进行全局优化提高分割阈值的求解速度,快速得到最佳分割阈值。该算法能够根据个体适应度大小和群体的分散程度自动调整遗传控制参数,从而能够在保持群体多样性的同时加快收敛速度,克服了基本遗传算法的收敛性差、易早熟问题。采用TI公司的DSP芯片TMS320VC5416,结合FPGA,搭建了多目标实时测量平台,并利用本文算法对图像阈值快速求解,实现了多目标的实时测量计算。实验结果表明,该算法具有良好的收敛速度和稳定性,阈值计算时间比二维Fisher准则法缩短了18ms(约63%左右);阈值范围稳定在3个像素以内,能够满足实时多目标测量要求。     

一种改进的模糊C-均值(FCM)彩色图像分割算法

针对传统模糊C-均值(FCM)聚类算法计算量大,聚类中心对初始值敏感和聚类数目不能自适应确定的缺点,提出了一种改进的FCM算法。首先对图像进行采样量化,并在满足视觉一致性的L*a*b*颜色空间计算并统计图像的色差信息,然后依据全局色差阈值选取初始聚类中心,对图像进行聚类分析,同时根据准则函数确定最佳聚类数,实现了聚类中心的优化选取和最优聚类数目的确定,有效减少了计算量。研究结果表明,改进后的FCM算法不仅较好地克服了传统FCM算法的缺点,而且聚类效果好,处理速度快,聚类效果与人的视觉感应保持了良好的一致性。     

单样本的低分辨率单目标人脸识别算法

针对只给定单幅目标图像的情况下,而要在监控视频中查找出该目标人脸图像的问题,提出了一种单样本的低分辨率单目标人脸识别算法。考虑到待识别样本集中的目标与非目标对象数量严重不均衡,以及单目标问题无法利用不同类别间的互斥关系。首先在待识别样本集中,通过聚类算法,将单目标的识别问题转化为多目标识别问题,进而提高开集人脸识别算法的鲁棒性;其次,利用迭代标签传播算法不断优化待识别样本的归属类别;在迭代过程中,按照置信概率估计每个类别的人脸确认阈值,以解决单样本无法训练分类器的问题。在多个人脸数据集上的实验结果表明,该算法对于单目标的单样本的人脸识别精确率既能逼近100%,也具有较高的召回率。     

Multiobjective image recognition algorithm in the fully automatic die bonder

It is a very important task to automatically fix the number of die in the image recognition system of a fully automatic die bonder. A multiobjective image recognition algorithm based on clustering Genetic Algorithm (GA), is proposed in this paper. In the evolutionary process of GA, a clustering method is provided that utilizes information from the template and the fitness landscape of the current population. The whole population is grouped into different niches by the clustering method. Experimental results demonstrated that the number of target images could be determined by the algorithm automatically, and multiple targets could be recognized at a time. As a result, time consumed by one image recognition is shortened, the performance of the image recognition system is improved, and the atomization of the system is fulfilled. __________ Translated from Chinese Journal of Scientific Instrument, 2005, 26(10) (in Chinese)     

带有炸点识别的图像采集系统设计

针对大范围炸点三维坐标图像测量过程中相机准确采集图像的问题,设计了可用于炸点识别的图像采集系统,以面阵相机采集到的图像数据为源信号,设计相应的算法自动识别炸点目标,从而为后续的图像采集及处理提供依据。设计了面阵相机图像采集的硬件,给出了炸点目标检测和识别的图像处理算法。实验验证表明,设计的算法可有效识别炸点图像中的火光、烟雾等信号。所研究的自动识别炸点目标图像采集算法和硬件,能减少相机存贮空间,提高靶场测试效率。     

基于遗传算法优化的切屑形态识别方法

将遗传算法用于图像模式识别,使用图像的参数模型将图像的模式识别转化为参数的局部优化。试验证明:利用遗传算法对神经网络权值进行优化,可提高识别的准确率和速度     

Clustering algorithm for texture data from remote sensing

A histogram-based clustering algorithm is proposed that takes into account features of the collection of image texture statistics. The algorithm allows the addition of the false clusters occurring on the boundaries of objects with different textures, thus significantly reducing their number. The clusters are analyzed by estimating their separability in the multidimensional vector space of features and the image context. The application of the algorithm to the automated recognition of types of land cover from aerial photographs of forest landscapes is considered. A comparison of cluster maps and schematic map of ground survey shows their good agreement     

Real-time cutting tool state recognition approach based on machining features in NC machining process of complex structural parts

Cutting tool state recognition plays an important role in ensuring the quality and efficiency of NC machining of complex structural parts, and it is quite especial and challengeable for complex structural parts with single-piece or small-batch production. In order to address this issue, this paper presents a real-time recognition approach of cutting tool state based on machining features. The sensitive parameters of monitored cutting force signals for different machining features are automatically extracted, and are associated with machining features in real time. A K-Means clustering algorithm is used to automatically classify the cutting tool states based on machining features, where the sensitive parameters of the monitoring signals together with the geometric and process information of machining features are used to construct the input vector of the K-Means clustering model. The experiment results show that the accuracy of the approach is above 95% and the approach can solve the real-time recognition of cutting tool states for complex structural parts with single-piece and small-batch production.     

基于聚类分析的说话人识别研究

通过分析当今说话人识别系统中常用的一些特征参数.提出一种基于改进的K均值聚类算法的多特征组合的说话人识别方法。经过多次实验证明.采用上述方法不仅解决了K均值算法对初始值敏感、易陷入局部最优的问题.而且有效地提高了系统的识别率。     

Algorithms for automated characterization of cell populations in thick specimens from 3-D confocal fluorescence microscopy data

Methods are presented for the automated, quantitative and three-dimensional (3-D) analysis of cell populations in thick, essentially intact tissue sections while maintaining intercell spatial relationships. This analysis replaces current manual methods which are tedious and subjective. The thick sample is imaged in three dimensions using a confocal scanning laser microscope. The stack of optical slices is processed by a 3-D segmentation algorithm that separates touching and overlapping structures using localization constraints. Adaptive data reduction is used to achieve computational efficiency. A hierarchical cluster analysis algorithm is used automatically to characterize the cell population by a variety of cell features. It allows automatic detection and characterization of patterns such as the 3-D spatial clustering of cells, and the relative distributions of cells of various sizes. It also permits the detection of structures that are much smaller, larger, brighter, darker, or differently shaped than the rest of the population. The overall method is demonstrated for a set of rat brain tissue sections that were labelled for tyrosine hydroxylase using fluorescein-conjugated antibodies. The automated system was verified by comparison with computer-assisted manual counts from the same image fields.     

A honeybee-mating approach for cluster analysis

Cluster analysis, which is the subject of active research in several fields, such as statistics, pattern recognition, machine learning, and data mining, is to partition a given set of data or objects into clusters. K-means is used as a popular clustering method due to its simplicity and high speed in clustering large datasets. However, K-means has two shortcomings. First, dependency on the initial state and convergence to local optima. The second is that global solutions of large problems cannot be found with reasonable amount of computation effort. In order to overcome local optima problem lots of studies done in clustering. Over the last decade, modeling the behavior of social insects, such as ants and bees, for the purpose of search and problem solving has been the context of the emerging area of swarm intelligence. Honeybees are among the most closely studied social insects. Honeybee mating may also be considered as a typical swarm-based approach to optimization, in which the search algorithm is inspired by the process of marriage in real honeybee. Neural networks algorithms are useful for clustering analysis in data mining. This study proposes a two-stage method, which first uses self-organizing feature maps (SOM) neural network to determine the number of clusters and then uses honeybee mating optimization algorithm based on K-means algorithm to find the final solution. We compared proposed algorithm with other heuristic algorithms in clustering, such as GA, SA, TS, and ACO, by implementing them on several well-known datasets. Our finding shows that the proposed algorithm works better than others. In order to further demonstration of the proposed approach’s capability, a real-world problem of an Internet bookstore market segmentation based on customer loyalty is employed.     

一种基于颜色聚类和种子填充的目标识别算法

针对类人足球机器人目标识别易受光照强度变化影响的问题,改进了常规的扫描线种子填充算法,提出了一种基于颜色聚类分割和种子填充的目标识别算法。该算法采用HSI颜色模型,依据亮度和饱和度信息对图像粗分割,基于阈值将彩色区域和灰色区域分离,同时将灰色区域二值化;基于色调直方图聚类分析对彩色区域的像素进行了归类,通过改进扫描线种子填充算法实现了色块的最简扩充和特征提取,使颜色分割和特征提取同步进行,解决了目标物体的快速、精确识别。实验结果表明,该算法抗干扰能力强,分割精度高,能满足实时性要求,具有一定的实用价值。