图像测量技术具有非接触性、抗干扰性等优点被越来越多地应用到双次级永磁直线电机动子位置检测中。针对图像采集过程中所产生的图像模糊问题，提出了一种精确辨识运动模糊参数的方法，实现了直线电机动子位置的精确测量。采用数字散斑图像作为原始拍摄面，通过相机获得包含动子位置信息的序列运动模糊图像。利用奇异值分解对运动模糊图像进行去噪重构预处理，降低噪声对运动模糊参数辨识精度的影响。求取预处理后图像的倒谱图，通过主成分分析找到倒谱图二值化后亮线的主矢量方向即为模糊方向；同时作出倒谱三维图，利用其负峰值中两个最小值坐标求出模糊长度。根据计算的模糊参数值复原模糊散斑图像，再利用图像亚像素测量方法获得动子位置。实验结果表明：本文算法的计算耗时比Radon算法缩短了5倍多，不同位移下的平均绝对误差仅为0.084 7 mm，满足直线电机动子位置检测的实时性和高精度要求。

Image measuring technique is more and more applied to the double secondary permanent magnet linear motor rotor position detection because of the advantages of non-contact and anti-interference. Aiming to the motion blur problem of image acquisition process, an accuration identification method of motion blur parameters was proposed to realize the precise measurement of linear electric motor rotor position. By using digital speckle image as the original photo surface, the motor rotor position information of sequence motion blurred image was obtained by camera. The singular value decomposition(SVD) was used to denoise the motion blur image for reducing the effect of noise on identification accuracy. The cepstrum image was calculated, and main direction of the cepstrum binary bright line, which was the blur direction, was found by using principal component analysis(PCA). At the same time, the cepstrum 3D image was made by using the two minimum values of its negative peak to calculate blur length. According to the parameters calculated, the blur speckle image was recovered, and rotor position was obtained by using sub-pixel image measurement method . Experimental results show that: the time consumption of proposed method is five times shorter than the Radon method. And the average absolute error under different dispplacements is only 0.084 7 mm, which satisfy the requirments of real time and high precision in linear electric motor rotor position measurement.Key words: