Abstract: Identifying vehicles in highway inspection is a critical task. In the proposed method, automatic real time vehicle detection is performed using adaptive background subtraction, Sobel filter, morphological operations for removing irrelevant objects and noise, and finally finding the centroid of remaining objects which are vehicles. Using these methods together provides excellent capabilities to detect and track vehicles in highway inspection. We consider the average total of 14 frames before the current frame as the background model. When a new object enters the frame, our algorithm detects it using the subtraction of background model from the current frame. Experimental results show the high accuracy of the proposed method with maintaining its low time complexity and Computational complexity.