Abstract: An improved method based on Zernike moment and Sigmoid fitting is presented in order to improve the weakness of low accuracy when Zernike moment edge detection method detects non-straight edges. This papwr uses Zernike moment edge detection method to detect edge, then uses Sigmoid fitting method to detect edge, and adopts weighted sum method to fix the detection result according to the results of two methods, the results are what we want. Both ideal images and actual industrial CT images are used to demonstrate its accuracy, the experimental results show that the detection precision of the method proposed has improved one order of magnitude comparing with Zernike moment edge detection method when detecting non-straight edges.

Key words: edge detection, Zernike moment, Sigmoid fitting, sub-pixel, weighted fixed

摘要： 为了改进Zernike矩边缘检测方法对非直线边缘检测精度低的弱点，提出了一种改进的基于Zernike矩方法和Sigmoid拟合法的边缘检测方法。用Zernike矩边缘检测方法检测边缘，然后用Sigmoid拟合方法检测边缘，根据边缘类型对两种方法检测的边缘结果进行加权求和，所得结果为边缘位置。分别利用理想图像和实际工业CT图像进行了实验验证，实验结果表明对非直线边缘的检测，该方法检测精度相对于Zernike矩方法提高了一个数量级。

关键词: 边缘检测, Zernike矩法, Sigmoid拟合法, 亚像素, 加权修正