This paper presents a height estimation method for video surveillance systems based on single view metrology. In the proposed method, a cubic box is used as the reference object for determining three rough vanishing points. An optimization problem is then formulated to fine tune these inaccurate vanishing points based on a novel object function and the genetic algorithm with Cauchy mutation operator. In video surveillance systems, the height of an object can be estimated by providing its bottom and top feature points. The advantages of the proposed approach are its independency of the camera's internal parameters and ease to setup. Experimental results show the effectiveness of the proposed approach with digital and analog video cameras.