A machine vision-based aperture measurement method was proposed to address the problems of large workload, slow speed, and subjective influence of corner connector aperture measurement. Firstly, the acquired image was greyed out, binarised, and converted to improve the image processing speed, locate the location of the corner connector quickly, and establish an accurate object-image relationship. Then, the Canny edge detection operator was used to extract the corner connector aperture contour at the pixel level to obtain a complex polygon contour. Finally, a caliper tool was built on the smallest outer circle of the complex contour to achieve sub-pixel precision extraction of the corner connector aperture edge. The experimental results show that the maximum measurement error of the method is less than 0.03 nm, the repeatability measurement accuracy is approximately 0.01 nm, and the measurement time of the system is 126 nm. Therefore, the method can meet the measurement accuracy requirements of the corner connector aperture and the measurement data has good stability; it can meet the needs of high-volume, high-intensity, real-time online inspection, and has good application prospects.