由于全聚焦成像技术主要集中于使用一维线阵实现二维全聚焦成像，文中基于二维面阵将全聚焦算法扩展至三维，并利用现场可编程门阵列的高速并行运算能力实现硬件的全聚焦过程，检测图像的刷新率高达20帧·s^-1，数据实时处理能力约为每秒2G字节，从真正意义上实现了实时三维全聚焦成像检测。通过仿真及系统试验表明，相比二维全聚焦成像结果，三维全聚焦成像结果更加直观，能够真实还原缺陷的整体结构。

The research of total focusing method (TFM) imaging technology mainly focuses on the use of one-dimensional array to realize two-dimensional total focusing method (TFM) imaging. In this paper, the focusing algorithm was extended to three-dimension based on two-dimensional array, TFM process was implemented by hardware based on high speed parallel computing ability of FPGA. The refresh rate of detection results image can be up to twenty frames per second, real-time data processing capacity is about 2G bytes per second, and hence the real time 3D-TFM detection is realized in real sense. The simulation and system experiments show that, compared with the 2D-TFM imaging results, 3D-TFM imaging results are more intuitive, and the whole structure of the defect can be truly restored.