Abstract: Cavity and delamination are common defects in brazed joints, and the quality of brazed joints was impacted by the area of defects. In order to calculate the defect area accurately and efficiently, the defect recognition algorithm was proposed to mark the defect. Firstly, ultrasonic microscopy system was used to detect the brazed workpiece and generate ultrasonic C-scan image. Then the C-scan image was binarized with the threshold-based image segmentation method. Finally, the defect was marked and the defect area was calculated using the proposed defect recognition algorithm. The accuracy and efficiency of defect detection was improved.