摘要
针对常规均衡类边界识别方法的识别结果发散的问题,在化极的基础上提出了基于曲率磁梯度张量矩阵的最大特征值的归一化改进算法,其零值等值线更接近磁源体的真实边界轮廓。同时,针对提取到的特征边界等值线连续性差的问题,采用最小二乘法对离散的等值点进行拟合。仿真及试验结果表明:最小二乘拟合结果为判断磁源体的中心位置、水平分布范围和大致走向等提供了定量参考,提高了边界识别结果的准确度。
Aiming at the problem of the divergence of the recognition results of the conventional balanced edge detection methods,an improved normalization algorithm based on the maximum eigenvalue of the curvature magnetic gradient tensor matrix is proposed on the basis of the reduction to the pole. The zero value contour of the proposed algorithm is closer to the true boundaries of magnetic bodies. Meanwhile,for the problem of poor continuity of extracted characteristic edge contours,the least square method is utilized to fit the discrete equivalent points. Simulation and experimental results prove that the least square fitting result provides a quantitative reference for judging the center position,the horizontal distribution range and the general direction of the magnetic bodies. The accuracy of the edge detection results is improved.
出处
《装甲兵工程学院学报》
2017年第5期113-118,共6页
Journal of Academy of Armored Force Engineering
关键词
磁源体
边界识别
曲率磁梯度张量
化极
最小二乘法
magnetic body
edge recognition
curvature magnetic gradient tensor
reduction to the pole
least square method
