Optical frequency-domain reflectance (OFDR) has been widely used in vibration measurement due to its unique advantages over optical time-domain reflectometry (OTDR). It should be noted that, however, OFDR requires long measurement time and shows poor sensitivity when applied to measure vibration signal over long distance. In the work, an algorithm is presented to automatically detect and locate the vibration signals. Firstly, we perform cross-correlation analysis in a moving window between the beating signals without and with vibration, and find the maximum cross-correlation coefficients in all windows to reconstruct them into a cross-correlation curve. Secondly, an automatic decision threshold curve is designed to conclude whether there is any vibration over the sensing fiber. Lastly, the cross-correlation curve is compared with the threshold to locate the vibration. We experimentally test the algorithm in an OFDR system and locate a PZT vibration at 26.96 km, which demonstrates its validity in terms of detecting external disturbances over a relative long distance.