基于改进Radon变换的直线钢轨识别算法

针对传统Radon变换在直线轨道检测方面速度慢且不能从众多边缘直线中识别出钢轨,提出基于改进Radon变换的直线轨道识别算法。首先,通过阈值筛选的边缘点;其次,利用先验知识,建立极角约束域;然后通过Radon变换提取所有直线,再利用直线角度差和直线平均距离筛选出钢轨边缘直线;最后将本文算法与传统Hough变换算法和传统Radon变换算法的比较,实验结果显示本文算法能精准识别出钢轨,且速度更快。     

剃齿刀刃磨技术

剃齿刀和其它金属切削刀具一样，切削工件数越多，其刀具切削刃磨损越严重，并导致切削工件几何精度下降，这时在加工件的尺寸超出它的允许公差之前，剃齿刀切削刃槽底出现磨损，切削刃变钝这时必须进行修磨，另外，阐述了剃齿刀直径与公法线关系及刃磨图表的应用，以及不同阶段的剃齿刀对齿轮齿形的影响及采取的相应措施。     

运动图像处理在汽车车型识别中的应用

本文介绍了“多品种混流机器人喷漆自动线的汽车车型识别系统”的一种算法,提出了基于运动序列图像的自适应阈值判别法和利用特征匹配法解决了同种车型的图像在结构上产生局部变化的识别问题。     

Real-Time Fuzzy Obstacle Avoidance Using Directional Visual Perception

This paper presents a novel vision-based obstacle avoidance approach for the Autonomous Mobile Robot (AMR) with a PanTilt-Zoom (PTZ) camera as its only sensing modality. The approach combines the morphological closing operation based on Sobel Edge Detection Operation and the (μ-kσ) thresholding technique to detect obstacles to soften the various lighting and ground floor effects. Both the morphology method and thresholding technique are computationaUy simple. The processing speed of the algorithm is fast enough to avoid some active obstacles. In addition, this approach takes into account the history obstacle effects on the current state. Fuzzy logic is used to control the behaviors of AMR as it navigates in the environment. All behaviors run concurrendy and generate motor response solely based on vision perception. A priority based on subsumption coordinator selects the most appropriate response to direct the AMR away from obstacles. Validation of the proposed approach is done on a Pioneer 1 mobile robot.     

基于Top-Hat-Canny的汽车前视摄像头图像边缘检测方法研究

针对汽车前视摄像头(Forward View Camera)拍摄存在噪声及光照不均匀的问题,提出一种基于顶帽算法TopHat Canny的多重边缘检测图像预处理方法。通过对前视摄像头拍摄的道路行人、车辆及障碍物的图片进行分析,采用小波阈值算法分别对彩色图像(Red-Green-Blue,RGB)3个空间分量进行滤波去噪,并经加权平均法对重组的3个分量灰度化处理后,采用Top-Hat-Canny多重边缘检测方法对图像进行处理。实验仿真结果表明:该方法可以有效地解决噪声及光照不均对图像边缘检测带来的影响,为后续目标识别工作提供基础。     

车牌定位的一种神经网络方法

汽车牌照自动识别（LPR）是ITS中的一项关键技术。对实时采集的图像，能自动地提取车牌照信息，有利于快速、高效进行车辆监控和管理，真正实现交通管理的智能化。而车牌定位又是车牌识别的必要的先决条件，直接影响LPR的成功与否。针对国标GA36—1992，结合图像的边缘特征，给出了一种快速可行的车牌照定位方法，供同行们参考。     

基于二次误差测度的带属性三角网格简化算法

给出一种基于边折叠和二次误差测度的快速简便的算法来简化带属性的网格模型。该算法通过分别建立几何和颜色属性二次误差测度来计算几何和颜色属性误差,用几何与颜色属性误差的总和来控制网格简化的顺序和精度。边折叠是根据某种误差测度将候选的边按照折叠代价排序,每次取代价最小的边进行折叠操作,直至满足给定的终止条件。二次误差测度采用点到平面距离的平方作为误差测度。应用实例表明,该算法既能保证简化模型同初始模型在几何上尽可能相似,又能较好地保留初始模型的颜色、纹理等属性信息。     

Stability Analysis of Long-Span Continuous Rigid Frame Bridge with Thin-Wall Pier

During cantilever cast in construction of high-pier and large-span continuous rigid frame bridges, structural stability in the longest cantilevered stage is very important. Based on a practical design case of a large-span continuous rigid frame bridge in Wuhan, the longest span stability coefficient is calculated with linear-buckling and nonlinear-buckling methods, respectively. The influences of both geometrical nonlinearity and the dual nonlinearity of material and geometry are considered. Numerical results indicate that the nonlinear solution is necessary to stability analysis because linear buckling loads axe much higher than those of nonlinear buckling. Thus, the edge fiber yield criterion is more convenient and faster than ultimate loading criterion when estimating nonlinear stability of structure, and can be used easily in the initial engineering design.     

基于Solid Edge装配环境下管道设计的研究与实现

在分析企业管道设计需求及Solid Edge软件的基础上，结合管道自身的特点，利用中心线进行管道的干涉判断与调整，提出了二次投影干涉检验算法。该算法基于发生干涉的物体共用相同的空间，其投影图形也必定相交的事实，将三维图形转变成二维图形进行干涉判断，大大地降低了计算难度。