Aiming at the problems of low efficiency, poor accuracy consistency, and reliance on empirical judgment in the manual dimension inspection of ceramic insulators during the production process, a sub-pixel-level visual inspection system based on the Halcon platform was designed. Taking the 95-porcelain insulators with a 60 × 60 specification as the research object, a three-layer inspection architecture of “hardware acquisition–software processing–data output” was constructed. Through key technologies such as camera calibration, distortion correction, sub-pixel contour extraction, and template matching, the automatic measurement of three core dimensions of the insulator, namely height, width, and shed distance, was achieved. The experimental results show that the detection error of this system is controlled within the range of 0.5–1.2mm, the detection success rate reaches 99.2%, the detection time per sample is  2s, and the efficiency is 40% higher than that of traditional manual inspection. It can accurately meet the dimension inspection requirements of “GB/T 772-2005 Technical Conditions for Porcelain Insulators for High-voltage Overhead Lines”. This system requires no human intervention, and the detection results are stable and reliable. It provides an efficient solution for the on-line quality control in the production process of ceramic insulators and has important engineering application value.