基于高斯积分曲面拟合的亚像素边缘定位算法

发布时间：2018-02-13 13:58

本文关键词： 视觉测量法截线过渡带高斯积分亚像素　出处：《仪器仪表学报》2017年01期 　论文类型：期刊论文

【摘要】：针对现有亚像素边缘定位算法存在精度不高、计算复杂的问题,提出一种基于高斯积分曲面拟合的亚像素边缘定位算法。根据单边阶跃状边缘特征,构建边缘法截线的高斯积分模型,在确定边缘过渡带的基础上,将拟合曲面区域内的像素点信息转化为边缘曲线的活动坐标,并对转化后的像素点坐标与灰度值按照高斯积分模型进行拟合,准确定位图像的亚像素边缘。采用所述视觉测量系统,用量块直线边缘进行实验,并与传统高斯曲面拟合亚像素边缘定位算法比较,证明基于高斯积分曲面拟合的亚像素边缘定位算法具有较高的定位精度,一等量块的直线度误差在1μm以内,计算速度提高一倍。采用该算法确定亚像素边缘时,可通过修正高斯积分模型的均值,有效补偿光源强度造成的误差。本算法可以应用于齿轮等高精度机械零件的测量。
[Abstract]:In order to solve the problem of low accuracy and complicated calculation in existing sub-pixel edge localization algorithms, a sub-pixel edge location algorithm based on Gao Si integral surface fitting is proposed. The Gao Si integral model of the edge section is constructed. On the basis of determining the transition zone of the edge, the pixel information in the fitting surface area is transformed into the active coordinate of the edge curve. The transformed pixel coordinates and gray values were fitted according to Gao Si integral model to locate the sub-pixel edges of the image accurately. Compared with the traditional Gao Si surface fitting sub-pixel edge location algorithm, it is proved that the sub-pixel edge location algorithm based on Gao Si integral surface fitting has high accuracy, and the straightness error of the first-class block is less than 1 渭 m. When the algorithm is used to determine the sub-pixel edge, the error caused by the intensity of light source can be effectively compensated by modifying the mean value of Gao Si's integral model. This algorithm can be applied to the measurement of high-precision mechanical parts such as gears.
【作者单位】：沈阳工业大学机械工程学院;沈阳机床股份有限公司齿轮分公司;
【基金】：国家科技支撑计划(2014BAF08B01)项目资助
【分类号】：TP391.41

【相似文献】