Real-time biscuit tile image segmentation method based on edge detection

In this paper we propose a novel real-time Biscuit Tile Segmentation (BTS) method for images from ceramic tile production line. BTS method is based on signal change detection and contour tracing with a main goal of separating tile pixels from background in images captured on the production line. Usually, human operators are visually inspecting and classifying produced ceramic tiles. Computer vision and image processing techniques can automate visual inspection process if they fulfill real-time requirements. Important step in this process is a real-time tile pixels segmentation. BTS method is implemented for parallel execution on a GPU device to satisfy the real-time constraints of tile production line. BTS method outperforms 2D threshold-based methods, 1D edge detection methods and contour-based methods. Proposed BTS method is in use in the biscuit tile production line.     

基于NSCT域边缘检测的侧扫声呐图像分割新方法

针对侧扫声呐图像混响噪声严重、对比度低、分割困难的问题,根据侧扫声呐图像成像特点,提出一种基于非下采样Contourlet变换(NSCT)域边缘检测和区域生长的侧扫声呐图像分割新方法.首先,对侧扫声呐图像进行NSCT分解,通过K均值聚类法在NSCT域低频部分对阴影进行分割;然后通过寻找NSCT域高频层由同一粗尺度分解的2个细尺度相邻子带系数差的模极大值位置,来选择图像边缘点,并进行尺度内、尺度间的边缘融合;最后,利用基于边缘的区域生长方法完成对目标的分割.实验结果表明,该方法对侧扫声呐图像分割具有抗噪性能好、正确分类率高以及边缘定位准确等优点.     

基于数学形态学的新方法在脑组织分割中的应用

针对人体脑部切片图像特点,提出了一种基于数学形态学的脑组织自动分割算法.该算法首先通过形态重构获得粗糙的脑组织区域,然后运用腐蚀和膨胀运算进行边界定位分割出了脑组织,最后对连续断层图像的分割结果进行了三维重建.结果表明该算法分割准确且自动化程度高,适合于大量序列切片图像的快速自动分割.     

采用局部分形的高效图像分割方法在红外云图处理中的应用

针对已有的基于分形维数的图像分割算法难以快速计算一个小区域的分形维数,计算复杂,效率低的问题,通过分析云的分形特征,提出一种采用局部分形维数的方法对红外云彩图像进行分割.首先,提出了一种高效算法来计算一段区间内的分形维数,使用树状数组作为数据结构,利用已经计算出来的信息,在O(logN)的时间内得到结果;然后,通过计算...     

基于区域生长法的自适应图像分割的网眼织物瑕疵检测

为了解决网眼织物在织造过程中的实时检测问题,提出了一种基于频域滤波和区域生长法的瑕疵区域自适应分割方法。通过分析傅里叶变换后的功率谱获得网眼织物纱线分布特征后,通过设计滤波器自动滤除了与织物纹理背景相关的频率成分以削弱织物自身纱线分布结构对瑕疵检测精度的干扰。利用滤波后图像的灰度分布特点自适应地求取了分割参数,并使用区域生长法实现了对瑕疵区域的自动分割。利用对获得的二值图像进行形态学运算,得到了最终的瑕疵区域分割图像。实验结果表明,该算法能较准确地对瑕疵进行定位,瑕疵区域分割精度较高,自适应性和抗干扰能力较强,可以满足工业检测的要求。     

Simulation of defects and stresses in castings

For cast aluminum wheel both the technology of low-pressure permanent mold casting and the bending test were simulated. The shrinkage defects and residual stresses were predicted by computer methods. The total stress fields were examined taking into account the metal structure and the technological features. Non-uniform strength distribution was detected in the cast alloy. It was shown that the technological residual stresses affect the magnitude and distribution of stress concentrators in the part under load.     

新的颜色相似度衡量方法在图像检索中的应用

相似度衡量与特征提取在图像检索中具有重要的作用.基于自适应颜色特征提取的技术,提出一种新的颜色相似度衡量方法,称作加权主色优先距离WMCF.它由3个视觉感知特性条件导出,据此改善图像检索效果.同时,也采用简化的脉冲耦合神经网络提取新的纹理特征,进一步提高图像检索的精确度.实验表明,新的相似度衡量方法相较于CHIC及OCCD衡量方法有着更高的精确度和较小的时间复杂度;同时结合颜色特征与纹理特征的最终检索方法相较FC、BDIP Nmi等方法在精确度上有10％左右的提高,且具有更好的相关图像排序结果.     

Application of edge detection method based on image quality gradient for twin detection

An image quality based method was applied to quantitatively analyse the contribution of electrodeposited copper twinning. This method, based on the image quality as obtained from electron backscattering diffraction, involves three processes: hexagonal edge detection, thinning algorithm and twin determination. The twin density determined by the new algorithm was 65.61% which is higher than the twin density, 57.44% as determined by the Orientation Imaging Microscopy (OIM) analysis. The newly developed algorithm provides an alternative to twin boundary detection.     

基于Otsu方法的钢轨图像分割

由于钢轨图像灰度分布不均,一般的图像分割法难以将目标从背景中分割出来,故本文提出了目标方差加权的类间方差阈值分割法对钢轨图像进行阈值分割。分析了钢轨图像的特点,总结了加权的目标方差(Otsu)方法及其它全局阈值分割法对钢轨图像分割存在的问题。然后,对Otsu方法进行改进,以目标出现的概率为权重,对类间方差的目标方差加权,使分割阈值靠近单模直方图的左边缘和双模直方图的谷底。最后,计算图像的错误分类误差、钢轨图像的缺陷检测率和误检率来验证算法的有效性。实验结果表明,改进的Otsu方法能有效地分割钢轨图像,错误分类误差接近0。与其它阈值分割法如Otsu法、其它改进的Otsu法、最大熵阈值分割法相比,本文方法对钢轨图像的分割效果更优,缺陷检测率和误检率分别为93%和6.4%,适合机器视觉缺陷检测的实时应用。     

埋弧焊X射线焊缝图像缺陷分割检测技术

针对在埋弧焊X射线焊缝图像的强噪声和弱对比度特点下常规图像分割算法成功率低的现状,通过综合的分析和实验有针对性地给出了系统、实用的缺陷分割方法.首先通过实验给出了一种提取焊缝图像感兴趣区域(the region of interest,ROI)的方法.该方法通过中值滤波,基于sin函数的图像增强、大津法分割、Sobel算子边缘检测和Hough变换可以定量计算出X射线焊缝图像的ROI区域.进一步通过实验给出基于大津法的焊缝缺陷分割算法,实验表明在无人工设定初始分割阈值情况下这一分割算法具有较高的分割成功率.为进一步提高分割成功率,针对焊缝缺陷相对面积较小的特点,提出将缺陷视为噪声,将平均局部平均灰度视为密度,利用密度聚类方法进行缺陷分割.该方法在78张有缺陷的焊缝图像中,成功地分割出74张图像中的缺陷.最后在所研究算法基础上给出了一个综合考虑到各种因素的分割算法,即可节约计算时间,又可以保证分割的成功率.     

一种CT图像的肺实质分割方法

为了实现计算机辅助诊断,解决自动化CT图像处理的问题,文中提出一种肺实质分割的算法.采用求直方图峰值间极小值的方法获取最优阈值,实现图像分割;在此基础上,采用空间滤波以及连通域标记法去除干扰得到肺实质区域.结果表明该方法能实现肺实质分割,解决了肺结节检测的预处理问题.     

基于提升小波的颤振图像边缘检测方法研究

这里基于提升小波理论重点研究了图像边缘检测问题，提出了一种新的图像多尺度边缘检测方法，该检测方法适用于多种线性或非线性双正交小波，具有运算速度快、检测精度高等特点；并基于此边缘检测技术，研究了飞机结构颤振的边界预测问题，给出了一种新的稳定性参数估计方法。     

适用于光照不均匀图像的边缘检测算法

首先利用图像梯度方向特征改进Canny算法,使之更适合连续边缘的检测。然后将其与基于侧抑制原理的图像增强算法结合,设计了一种新的适用于噪声背景下光照不均匀图像的边缘检测算法。算法实验结果表明该边缘检测算法能有效提取光照不均匀图像中的连续边缘。     

模糊C均值算法在烧结断面图像分割中的应用

利用模糊C均值的算法,可以实现烧结图像的自动分割。传统的模糊C均值算法在判定像素归属时没有考虑像素的空间信息,对在烧结图像上存在的大量噪声非常敏感。为提高烧结图像分割的准确性,对传统FCM算法中的目标函数进行修正。实验表明采用FCM算法对烧结图像有良好的分割效果,而改进的FCM算法对图像中噪声具有较强的鲁棒性。     

基于改进型PCNN图像分割技术的金属表面检测

对金属表面缺陷中的轧入氧化皮、斑点和划痕3种缺陷检测设计了一种有效方法.首先将目标图片进行灰度值归一化增强图像,提高其视觉效果方便后续图像处理.然后应用改进型脉冲耦合神经网络图像分割技术将目标图片分割得到二值图像,因为三种缺陷各有其独有的特征,据此将二值图像进行分析对比,最后对已经检测出的三种缺陷采取不同的方法标注缺陷...     

Application of an improved watershed algorithm in welding image segmentation

In view of the present state of welding flaw X-ray test method and open question, a watershed segmentation method is proposed based on the dynamic combination rule. In consideration of image structure information, this paper presents an improved method based on the watershed algorithm, which has effectively restrained the over-segmentation occurrence by means of combination whilst segmentation in accordance with the dynamics combination rules. Experimental results show that the algorithm can quickly and accurately obtain the segmentation result of flaw image. Furthermore, it has higher ability in resisting noise.     

基于多路径网络的权值调整图像语义分割算法

为提高图像语义分割准确程度,针对场景解析中类别边缘分辨清晰度,提出了一种基于多路径网络的权值调整图像语义分割算法。通过引入多路径网络和权值调整并对图像场景中的物体类别具有的特征进行分析,提高图像的语义分割的准确程度;通过采用ADE20K数据集进行训练,提高边缘信息的分割效果,使模型具有更好的泛化能力。此算法加快了网络收敛速度。     

基于区域生长的生物图像分割方法

针对基于区域的分水岭算法中通常存在的严重过分割现象,采用两点措施来改进:其一是在分水岭分割之前采用迭代阈值分割算法对图像进行分割,从而有效地抑制了由图片表面的灰度变化引起的过分割,使边缘定位更加准确;其二是在分水岭分割之后以基于面积,对比度和深度控制的区域合并准则,有效抑制过分割现象,同时满足感兴趣区的分割要求,实验表明该方法简单有效,能够得到符合人类视觉系统特性的分割结果。     

基于多小波的图像处理在电晕检测中的应用

提出了基于多小波变换的图像处理方法,该方法以多小波变换为基础,在一次多小波分解与重构之间完成双谱段图像处理.首先进行多小波变换,将变换系数进行软阈值收缩消去噪声;然后根据图像中需增强的信息,选择增强系数进行子带增强;最后提出一种新的自适应权值融合规则,采用这个规则融合变换系数,进行小波重构得到处理后的单幅图像.实验表明,这种方法不仅能提高图像的视觉效果,增强源图像的边缘信息,而且能很好地将源图像中列电晕检测有用的信息融合在一起,提高电晕检测系统的定位精度.     

Prevention of defects in castings using back propagation neural networks

Defects in castings often lead to rejection, which would ultimately result in loss of productivity for a foundry. Expert systems developed by some researchers mostly act as postmortem tools, discussing and analyzing a defect after it has occurred. Though some investigators have attempted to predict a few important defects, a tool that could predict all the possible defects just before the castings are made has not yet been developed. Hence in the present work, an attempt has been made to predict major casting defects like cracks, misruns, scabs, blowholes and air-locks using back-propagation neural networks from the data collected from a steel foundry. The neural network was trained with parameters like green compression strength (GCS), green shear strength (GSS), permeability, moisture percent, composition of the charge and melting conditions as inputs and the presence/absence of defects as outputs. After the training was over, the set of inputs of the casting that is going to be made was fed to the network and the network could predict whether the casting would be sound or defective. If defective, the nature of the defect was also specified by the neural network. The neural network could predict cracks, misruns and air-locks accurately in most of the cases. The neural network could also predict other defects successfully. Investigating the causes followed by altering the moulding parameters and appropriate treatment of the molten metal can prevent the defects that were predicted by the backpropagation neural network.