基于图像识别技术在变电站设备监测中的应用

为减少传统基于灰度的模板匹配算法的计算量,采用一种在图像匹配中基于灰度的编码方式,将待搜索的源图像分割成若干块大小相同的子图像,计算每个子图像中的总灰度值,然后根据该子图像与相邻的子图像灰度值的排列关系进行编码,最后对每个子图像的编码值进行比较,实现图像的匹配。在图像匹配过程中,只需对编码值进行相等比较,并且每个子图像的编码计算简单,减少了传统模板匹配算法的计算量。仿真结果表明,该匹配算法能快速、准确地对变压器进行定位识别。     

一种具有旋转不变性的模板匹配方法

对目前应用的旋转图像进行模板匹配定位问题，通常采用由不同旋转角度的模板构成模板集合，用集合中的成员分别作用于目标图像进行匹配定位，该方法计算量大，很难将其应用于实时系统中。针对上述存在的问题，文中提出了一种借助非旋转模板对平面内旋转图像进行匹配定位的方法，第一步采用圆投影法对可能出现的匹配点进行预筛选，再用具有旋转不变性的NMI特征对筛选出来的点作进一步判断确定是否为存在模板图像的点。实验结果表明：该方法不但具有旋转不变形，而且可提高匹配定位的速度。     

图像识别技术在电力设备监测中的应用

为减少传统基于灰度的匹配算法的计算量,采用图像形状特征分析技术,提出了新的模板匹配算法。该算法取隔行隔列的数据进行粗略匹配,在此基础上进行精确匹配,大大减少了计算的数据量。将该算法应用于电力变压器的图像识别中,实验结果表明该匹配算法在速度上具有较大的优势,能更快地对电力变压器进行定位识别。论述了远程数字视频监控与图像识别技术在电力系统中的应用前景。     

图像识别技术在电力设备监测中的应用

为减少传统基于灰度的匹配算法的计算量,采用图像形状特征分析技术,提出了新的模板匹配算法.该算法取隔行隔列的数据进行粗略匹配,在此基础上进行精确匹配,大大减少了计算的数据量.将该算法应用于电力变压器的图像识别中,实验结果表明该匹配算法在速度上具有较大的优势,能更快地对电力变压器进行定位识别.论述了远程数字视频监控与图像识别技术在电力系统中的应用前景.     

基于模板匹配技术的航空发动机矢量作动筒标定研究

针对当前我国航空发动机矢量作动筒标定过程中,存在依靠人工游标卡尺测量效率低的问题,提出了基于机器视觉模板匹配技术的航空发动机矢量作动筒标定方法。该方法采用工业摄像头动态实时采集作动筒图像,通过对图像进行模板匹配设定RIO区域达到对作动筒长度的非接触测量。实践证明采用该方法进行航空发动机作动筒标定具有较高精度的同时还能有效提高标定效率。     

思维进化计算在图像识别中的应用

本文对思维进化计算(mindevolutionarycomputation,MEC)在图像识别中的应用进行了验证,结合模板匹配对多目标图像进行搜索。采用的方法是将模板和图像从RGB彩色空间转化到L×u×v空间进行匹配,通过MEC计算评价函数找到最优目标,最后对差图像进行分析。匹配结果表明,MEC具有平移、旋转和尺度不变的特性,在图像识别过程中取得了满意的效果。     

基于机器视觉的油画棒检测系统的设计

油画棒装盒传统方式采用人工,针对其摆放效率低、成本高的问题,设计了基于机器视觉的油画棒装盒系统,选用ARM芯片做图像处理,在有限的资源配置和时间条件下完成任务.系统通过摄像头对油画棒进行图像采集,控制电机转动油画棒,进行图像匹配,完成投放.系统先对图像定位、选取,滤波去噪,再采用改进的自适应阈值的直方图匹配算法,并结合模板图像和待匹配图像的相关系数,有效降低了计算量,提高了系统的识别率.实践证明系统的工作稳定,能实现对油画棒角度的有效调整,匹配速度快,单台识别率达到99％以上,能完成产品的自动装配.     

基于图像匹配的输电导线舞动监测方法

输电导线舞动对电网的安全运行造成了严重威胁。为此,提出了基于图像匹配的输电导线舞动监测方法。通过输电导线视频监控装置获取导线舞动的视频信息,以输电导线舞动视频中截取的多帧连续图像为研究对象,选取导线间隔棒作为跟踪目标来创建轮廓特征模板,通过匹配导线间隔棒来定位导线位置和计算舞动偏移量。该方法首先提取出未发生舞动时导线上的间隔棒进行模板创建;再对输入的待分析图像进行预处理,从而确定模板待匹配区域;最后通过后续视频帧图像中的全图搜索进行轮廓特征匹配,通过匹配成功的导线间隔棒来确定导线位置。实例分析结果表明,该方法能够匹配跟踪到导线舞动位置,并计算出导线舞动的相对偏移量和偏移角度。     

遗传算法在机器人视觉伺服控制中的应用

在机器人视觉伺服控制的研究中,对目标的识别是视觉伺服控制中的关键点,而目前图像目标的识别方法有模板匹配法和特征匹配法,计算量都比较大,本文采用基于全局和局部遗传算法相结合,提高目标识别的精度和速度,优化视觉伺服系统的性能。     

一种基于视频图像的运动车辆跟踪方法

基于视频图像的车辆运动跟踪是当前计算机视觉研究的热点,具有广泛的应用领域。本文研究了一种基于视频图像的运动车辆跟踪算法,通过当前帧目标边缘与实时更新模板的最优匹配来确定目标的位移量。首先启动了卡尔曼滤波,预测目标匹配搜索区域,然后再在搜索区域中利用边缘匹配精确定位目标,减小了匹配的计算量。实验结果表明,该方法是有效的且实时性好。     

Rotation invariant image matching by interpolated eight‐direction block matching

Template image matching is a method for seeking a target image area that is similar to a registered template image. However, template matching is a general‐purpose technique for various image applications, and the image under consideration must generally have the right position with respect to the template image. Therefore, the computational cost is large when both the location and inclination of the target image area are unknown. This paper proposes a method of searching for a target image area with arbitrary location and inclination by multiple matching using separated block template images. In the method, the template image is divided into small block areas with a size of 3 × 3 pixels. The image pattern in each block is rotated by 45^° and matching is performed between the rotated blocks and the image under consideration. The experimental results show that the proposed method has higher matching reliability than the conventional methods. © 2005 Wiley Periodicals, Inc. Electr Eng Jpn, 154(1): 66–73, 2006; Published online in Wiley InterScience ( www.interscience.wiley.com ). DOI 10.1002/eej.20107     

Rotation‐invariant image matching by geometrical parameter of equidistant edge pixel group

Template matching is a method of searching for a target image area that is similar to a template image. However, the target image area generally must have the right position relative to the template image. Therefore, the computational cost is high when the location and the inclination of the target image area are unknown. This paper proposes a method of searching for a target image area with arbitrary location and an arbitrary inclination by means of geometrical parameters obtained by an equidistant edge group. In this method, the difference of directions is calculated from the edge direction of a referenced edge pixel and from the directions from the referenced edge pixel to other selected edge pixels. After that, matching is performed, using the difference in directions and the matching rates are accumulated. Then the location and inclination of the target image area are sought by scanning the accumulation space. Experimental results show that the proposed method has higher reliability and lower computational cost for image matching. © 2007 Wiley Periodicals, Inc. Electr Eng Jpn, 161(1): 34–42, 2007; Published online in Wiley InterScience ( www.interscience.wiley.com ). DOI 10.1002/eej.20520     

快速区域质心图像匹配算法

针对传统的归一化互相关方法（NCC）在图像匹配时计算量大的缺陷,提出了一种新的基于NCC的快速模板匹配算法。该算法基于粗、精匹配的思想,将区域质心特征、颜色直方图信息与归一化互相关匹配方法结合起来实现图像的定位匹配。在图像匹配时以搜索图像的质心点作为参考点,而非通常情况下以左上角点作为参考点,减少模板的搜寻空间。通过大量的实验表明,该算法可以大大提高全搜索NCC匹配算法的运算速度,同时保持了匹配的精度,可以满足实时性的要求。     

Adaptive‐search template matching technique based on vehicle acceleration for monocular visual odometry system

One of the important tasks of an autonomous mobile vehicle is the reliable and fast estimation of its position over time. This paper presents the development of an adaptive technique to hasten and improve the quality of correlation‐based template matching for monocular visual odometry systems that estimate the relative motion of ground vehicles in low‐textured environments. Moreover, the factors that can affect the maximum permissible vehicle driving speed were determined and the related equations were derived. The developed system uses a single downward‐facing monocular camera installed at an optimum location to avoid the negative effect of directional sunlight and shadows which can disturb the correlation. In addition, the normalized cross‐correlation method is implemented to calculate the pixel displacement between image frames. Although this method is highly effective for template matching because of its invariance to linear brightness and contrast variations, it incurs high computational cost. Thus, the optimal sizes of image template and matching search area are selected and their locations are dynamically changed according to vehicle acceleration, in order to achieve a compromise between the performance and the computational cost of correlation. The proposed technique increases the allowable vehicle driving speed and reduces the probability of template false‐matching. Moreover, compared to traditional full search matching techniques, the adaptive technique demonstrates high efficiency and accuracy and improves the quality and speed of the correlation with more than 87% of reduction in computational cost. © 2016 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.     

基于图像匹配的电路板故障检测与定位算法

自动光学检查是利用图像处理算法进行电路板故障检测与定位的技术,它可以解决传统电路板测试无法克服的缺陷.图像相减是自动光学检查中最简单和最直接的办法,但具有一定的局限性,这是由于二值化预处理造成的.本文提出一种基于图像匹配的加载电路板故障检测与定位算法,无需进行二值化预处理操作,利用图像差的绝对值、平方以及相关操作,检测与定位元件的倒装与漏装故障,并具有与图像相减类似的简易性,实验证实了该算法的性能.     

一种基于图像特征点的图像匹配算法

图像匹配技术被广泛用于人脸识别、全景图像生成等领域.该文利用变比不变特征点 (Scale Invariance Feature Transform-SIFT)提取方法提取特征点,并对SIFT方法提取出的特征点用最近邻算法 (Nearest Neighbor-NN)进行匹配,在搜索最近邻特征点和次近邻特征点时使用了在K-D树搜索算法基础上进行改进的搜索算法BBF(Best Bin First)算法.实验证明该匹配算法具有匹配精度高,鲁棒性好的特点.     

改进 MOSSE 的小面积滑动指纹图像追踪算法

随着指纹传感器采集的指纹图像趋向于小型化,指纹图像所包含的指纹特征信息越来越少。针对传统模板匹配算法在处理小面积滑动指纹时计算量大、精度不理想、抗干扰能力差等问题,本文提出一种基于MOSSE的改进滑动指纹追踪算法。改进MOSSE算法使用多输入,将灰度特征与HOG特征在响应层加权融合,并引入Fourier-Mellin算法、加汉宁窗用以处理发生旋转的指纹。通过多种算法对小面积指纹进行跟踪的结果进行对比,表明本算法继承了原MOSSE算法的优点,并提高了指纹匹配精度,对正常图像匹配精度为99%,对含噪声图像匹配精度为90.3%,每帧均值计算时间为0.103 6 s,保证了指纹追踪的实时性,鲁棒性强,对产生形变和旋转的指纹图像也能进行很好的跟踪。     

对动态目标进行匹配的快速算法

基于图像匹配模型,在小波变换和投影特征的基础上,提出对动态目标进行匹配的快速算法.首先对动态目标进入视场的方向进行划分,针对划分后的区域利用小波变换和垂直投影技术得到一个可能的匹配点集合,再将模板图在搜索图上对匹配点集合中对应的点上利用灰度相关归一化公式作匹配运算,从而得到匹配位置.实验结果表明,该算法在实现匹配精度的同时,更进一步的提高匹配速度.     

Facial identification system based on a high‐speed matching algorithm for isodensity lines

In this paper, a high‐speed personal identification system which is adaptable for slant faces is presented. In the proposed method, the 25 facial images which are taken with different angles are registered in the database as reference images. The similarity of an input image is first examined between all registered images and the three facial images, which are similar to the input image, are then extracted as the candidate facial images. The isodensity map, which is expected to have the maximum similarity to the input image, is created by the interpolation between two candidate images and the input facial image is judged identical or not with the interpolated images. A template matching is adopted to obtain a similarity between two isodensity lines. In general, template matching requires a lot of time to obtain maximum similarity, while on the other hand it is very simple to use. To reduce the processing time for the matching efficiently, the techniques which are suited for the line pictures and the Sequential Similarity Detection Algorithm (SSDA) are adopted. Computer simulation using 50 people shows an accuracy of 95.2% correct discrimination. In addition, processing time is reduced to approximately 1/12. From the experimental results, the prospects of using this method are very encouraging. © 2003 Wiley Periodicals, Inc. Electr Eng Jpn, 143(4): 31–41, 2003; Published online in Wiley InterScience ( www.interscience.wiley.com ). DOI 10.1002/eej.10147