<p>In the intelligent welding process, it is difficult to obtain the penetration state of the weld seam by directly measuring the weld width inside the container, and it is difficult to obtain the welding defects in time. In order to improve the effect of rapid detection of welding defects, this paper combines intelligent machine recognition technology to study the rapid detection of welding defects. This paper proposes a pyramid hierarchical search strategy of moment feature matching method to accurately match and locate the welding area, and introduces image pyramid to reduce the amount of computation, and designs a search method of rough positioning at the top of the pyramid and then accurate positioning in the original image to meet the real-time requirements. Moreover, this paper uses image enhancement method to eliminate the interference of flux impurities to enhance the edge contour information and improve the accuracy of subsequent contour extraction. In addition, this paper combines with experiments to verify the effectiveness of the proposed method for rapid detection of welding defects. By comparing the experimental results, it can be seen that the intelligent machine recognition technology proposed in this paper can play an important role in welding defects.</p>