The storage of hazardous chemicals is related to the safety of people and city. At present, a method that hazardous chemicals are stored in the warehouses of the bungalows is adopted widely. The monitoring of all safety status are key problems, especially for the safety distance measurement of “five-distance”. In this thesis, aiming at the problems of effective corners extraction in binocular visual distance measurement, a united algorithm that based on Shi-Tomasi corners detection and improved Piecewise straight-line fitting is presented, in which a method of removing noise corners by using K-Nearest Neighbors algorithm is used and effective corners are intersections of fitting line. The experimental results show that the method can effectively remove false corners and recover the missing effective corners. The accuracy of corner extraction is more than 80%, and the distance measurement error is less than 2 cm.