提出了一种基于A扫描的滑动中值滤波算法，其通过对被检测工件与圆形超声直探头声场进行分析来确定滑动中值滤波的采样点数。对空心车轴超声检测系统采集的A扫描数据进行算法处理后可以发现，由算法处理后的A扫数据还原而来的B扫图、C扫图干净清晰，一方面去除了所处环境、机械运转结构及装配等因素造成的随机干扰；另一方面很好地保留了缺陷信息。

In this paper, a sliding median filtering algorithm based on A-scan was proposed, and the sampling points of the sliding median filter were determined by the analysis of the sound field of the circular ultrasonic probe and the actual workpiece.The B-scan and C-scan, which were restored by the A-scan data collected by ultrasonic inspection of hollow axles and processed by the algorithm, were clean and clear. On the one hand, this algorithm removed the random interference caused by the environment, mechanical structure and assembly, whereas on the other hand, it also retained the defect information.