Car detection is considered one of the best solution that used to reduce the congestion problem in big cities. There are many attempts to solve the problem based on Millimeterwave radar, and magnetic loop sensor. However, car delectation based on video camera is more robust because it is more accurate and gives information can be easily understood by human. In this work, we propose methodology for car detection in outdoor environment. Our method integrates HOG features and SVM to determines whether there is a car or not in the captured frame. Extensive experiments were performed by changing SVM parameters to gain 99% of successful classification. Moreover, we develop a new database with multiple environment noise, hard conditions and proved methodology in another dataset[13] that reached similar results.