摘要
针对局部灰度特征方法在高噪声图像中定位精度低的问题,提出改进局部灰度特征方法并用于工件亚像素边缘检测。首先,使用改进的Canny算子代替基于一阶导数的梯度算子,以便更精确地提取粗边缘;然后在采集到的像素窗口两侧建立子图像,代替单个窗口特征以修改边缘上下两侧强度值,再将新构造的子图像中间列像素进行加权,加速图像迭代复原速度;最后,通过所有新生成的子图像局部灰度特征重新计算亚像素位置,进一步提高检测精度。实验结果表明,在无外加噪声和加入1.00%、5.00%的噪声标准偏差图像中,RMS误差明显降低,亚像素坐标数与定位精度明显提高。
In order to solve the problem of low localization accuracy of local grayscale feature method in high noise image,an improved local grayscale feature method is proposed and applied to the sub-pixel edge detection of workpiece. First of all,using the improved canny operator instead of the traditional derivative mask to extract rough edge more precisely,and then collected the pixel window on both sides to establish the image instead of a single window feature to modify the edge strength value on both sides. Then,the intermedi-ate column pixels of the newly constructed subimage are weighted to speed up the iterative recovery of the image. Finally,the detection accuracy is further improved by recalculating the sub-pixel position through all newly generated sub-image features. The experimental results show that the RMS error is significantly reduced in the absence of additional noise and the addition of 1.00% and 5.00% noise standard deviation images.
作者
张智凡
于凤芹
ZHANG Zhi-fan;YU Feng-qin(College of Internet of Things Engineering,Jiangnan University,Wuxi 214122,China)
出处
《软件导刊》
2019年第6期158-162,共5页
Software Guide
基金
国家自然科学基金项目(61573168)
中央高校基本科研项目(JUSRP51733B)
关键词
边缘检测
亚像素
局部灰度
图像强度
迭代复原
子图像特征
edge detection
sub-pixel
partial gray features
image intensity
restoration iteration
subimage feature
