Abstract: In view of the current software security test algorithm's poor accuracy, outlier location accuracy and low detection efficiency, a software safety test algorithm based on feature point matching is proposed to complete the software safety test. Determined by matrix factors and factors, and the matching index matrix and attribute weight vector calculation, calculation results are given for similarity; similar instance rules are determined to obtain high precision software anomaly detection results; locate the outliers by two method, and converts it to a binary vector of multiple anomalies; the binary vector is calculated to get the software abnormal location set. The experimental results show that the proposed algorithm not only improves the accuracy of detection and location of the software outliers, but also improves the efficiency of software security detection.