This study proposes a novel subpixel edge extraction algorithm for the detection of defects in workpieces, which is based on Zernike moments. First, the target image is decomposed using a wavelet transform, and the decomposed frequency information is preprocessed by employing different algorithms. After reconstruction, the image noise can be effectively filtered out and the target information can be enhanced. Then, the proposed subpixel edge extraction algorithm is applied to locate the image edges and extract their feature information with the aim to reduce the edge information error and segment the target contour more accurately. Finally, the geometric parameters of the surrounding region and the global information entropy are calculated to determine whether there are defects. The algorithm is verified with an experiment, and the experimental results show that the proposed algorithm can reduce the metal high-light noise and extract the defect edges effectively. Moreover, the algorithm is robust even when the ambient light illumination changes, and thus improves the accuracy of metal-defect detection.