首页 > , Vol. , Issue () : -
As an important earthquake precursor, the anomaly from infrared remote sensing images needs stable and effective extraction algorithms. However, many algorithms have been only verified in a few earthquake cases and lack statistical analysis of abnormal signals. In this paper, a method is proposed to estimate the accuracy and universality of the algorithm through statistical analysis of abnormal signals from the power spectrum of infrared remote sensing images in China and surrounding areas. The method not only analyzes the amplitude, coverage of abnormal signals and relevant seismic information but also regards continuous abnormal points in time and space as a sample to calculate the positive predictive value of abnormal signals and the true positive rate of earthquakes with different parameters. Abnormal signals were extracted from FY-2G satellite infrared remote sensing images and statistically analyzed, which get the positive predictive value of 20.37% and the true positive rate of 65.96%. The abnormal signals with high amplitude and wide area have a positive predictive value of 80%. For earthquakes with a magnitude larger than 5.4, the true positive rate can reach 81.82%. This method can be used to analyze the characteristics of abnormal signals and evaluate the correlation between abnormal signals and earthquakes and be beneficial to the comparison and improvement of the algorithm.