超声相控阵图像的缺陷处理目前停留在人眼观察和人工标识的阶段。提出了一种自动识别超声相控阵缺陷图像的算法过程。该算法分为预处理、图像滤波、特征提取和多目标缺陷跟踪等步骤。通过分析目标缺陷轨迹的相关参数，结合目标缺陷自身的面积等信息来综合判断目标缺陷是否为实际缺陷。对带孔金属铝试块进行的测试检验了算法和程序的正确性，获得了很好的缺陷跟踪和识别效果。

Ultrasound phased array image process and analysis of the defect still relies on the human eye observation and artificial identification. A scheme for automatically processing and analyzing ultrasonic phased array defective images was proposed. The process is composed of preprocessing, image filtering, feature extraction and multi target defect tracking. By analyzing the relevant parameters of the target defect trajectory and combining the area of the target defect, it is determined whether the targeted one is a real defect. The accuracy of the algorithm and program is determined by the test of metal aluminum specimen with holes,and good effect of defect tracking and recognition is obtained.