边缘定位是锂电池涂布测量的关键，但由于生产的震动和纠偏原因，边缘会出现虚化和卷边等现象。为了提高锂电池涂布在线检测的精度和鲁棒性，本文结合锂电池涂布生产的特点提出了边界渐进定位法，实现了边缘的快速准确检测。首先利用跨邻域方向边界检测算子进行初次边缘定位，以提高弱边界的检测能力；再利用局部极值差分算法配合选择性峰值排序获得边缘像素级定位，保证了边缘投影的保峰性和抗噪性，提高了边缘定位的效率；最后，基于分段三次样条边界插值和分段直线拟合方法，实现了边缘的亚像素高精度定位。实测结果表明，算法在直线型边缘上具有较强鲁棒性。

Edge detection is a key step for the online vision measurement of lithium battery coating (LBC). However, as the vibration and rectification in LBC production, the virtualization and curling of edges could occur. In order to improve the accuracy and efficiency of on-line measurement of LBC, this paper proposed a staged edge location method according to the production characteristics of LBC, and achieved the swift and accurate detection of edges. Firstly, a cross neighborhood operator is used to detect the edge preliminarily to improve the ability of week edge detection. Then, local extreme values difference (LEVD) algorithm with selective peak sort algorithm is proposed to guarantee for the ability of edge-preserving and anti-noise of edge projection and to improve the efficiency of edge detection. Finally, piecewise cubic spline interpolation combined with segmented linear fitting method is provided to realize the sub-pixel location of the edge. The experimental results show the effectiveness of the method.