Using Haar-Like feature extraction shown as Eq. (7) to calculate segmentation blocks feature parameter as show in Figure 9. Therefore, make the car of bottom of the shadow is black blocks and white block for the up, right and left three directions respectively. And then segmentation vehicle block as show in Figure 10 from original image as shown in Figure 11 and calculate to gray value of sum of each block, and then each block of sum to normalize. The normalize value is the vehicle feature, if the vertical feature value is more than 0.15 and horizontal feature value more than 0.4, then marking the vehicle.