Abstract: In order to achieve spatial locating and visualization of batch defects in fillet weld of T joint weldments, detection images of weldments were obtained by using multi-view imaging method, and mathematic models of defect depth and deviation were established. On the basis of defects segmentation of detection images, according to the proposed automatic corresponding criterion of defects and automatic extraction algorithm of projection distance, the automatic extraction of spatial position data and visualization of batch defects in the joints were achieved. The data of 84 defects were analyzed statistically, and the distribution feature of defects was obtained. The experimental results show that the defects are mainly located on the left side of the sternum center line, which account for 73.8 percent of the total. The defects depth is mostly among 1.8 2.5 mm, which account for 85.7 percent of the total. Nearly 86 percent of the defects are mainly distributed in the joint surface of the wing plate and sternum. The defects radius is among 0 0.51 mm. The defects distribution is random alonging the weld longitudinal direction.