A stripe skeleton-level sub-calibration algorithm was used for interference fringe detection. The experimental samples of laser interference fringe were processed. An adaptive window was used to filter the interference fringes of different shapes, and the experimental samples were denoised to obtain the interference images with high contrast. The interference fringe extraction was performed according to the perfect sample images, and the teal-time stripe calibration and detection were then performed. The results show that the proposed multi-noise-based interference fringe detection model can realize the real-time detection of optical components and effectively improve their detection efficiency. Moreover it can realize the industrial automation detection based on the C++ platform.