In this article, a new approach to explore the candidate car license plate scene is presented. First, the image is decomposed by using a Haar wavelet to get the HL sub-image with vertical edges. Next, a black top-hat algorithm is applied to minimize the effects of interfering large continuous features other than the license plate. Then, the horizontal and vertical projections of the HL sub-image provide an approximate location of the license plate. The experimental results show the high precision of the proposed technique, with a success rate of 98% for parked vehicles and 97% for moving vehicles.