Abstract: The position and morphology of shock waves are critical indicators for evaluating the aerodynamic performance of key components in hypersonic vehicles. Quantitative extraction of shock wave features is essential for vehicle design optimization. However, in schlieren imaging of supersonic flow fields, high background noise caused by illumination conditions and window materials severely degrades the accuracy of shock wave identification and rapid feature extraction. Under high-noise conditions, the gradient magnitudes of background and shock waves are often similar, making conventional image processing techniques (e.g., filtering, frequency-domain transforms, and contour extraction) ineffective in distinguishing and suppressing noise, and thereby hindering accurate shock wave extraction. To address this issue, this paper proposes an automatic shock wave detection method based on the Line Segment Detection (LSD) algorithm combined with angle-based statistical filtering. The proposed method first eliminates invalid image regions through threshold analysis, followed by Gaussian filtering and region exclusion for preprocessing. The LSD algorithm is then applied to detect line segments, and the most representative shock wave segments are selected based on angle distribution statistics, with the highest-frequency angle range given priority, as well as segment length and spatial position. Experimental results demonstrate that the proposed method accurately identifies primary shock structures and extracts quantitative features in both high-noise (signal-to-noise ratio is 5.23 dB) and low-noise (signal-to-noise ratio is 20.38 dB) images, achieving a batch processing speed of 50 frames per second. The method exhibits strong robustness and real-time performance, offering practical value for shock wave analysis in hypersonic applications.