Optimized HOG Descriptor for on Road Cars Detection

The need for road security is increasingly highlighted, especially with the increase in road accidents around the world. One of the major causes is the vehicle speed and the drowsiness of drivers. Recently computer vision has become a well-developed research area especially in term of objects detection and tracking. Then, the exploitation of these developments for the road safety profile is strongly encouraged. Here we present our contribution to detect moving vehicles using optimized HOG process based on shape and motion parameters fusion. Indeed we prove in this paper that HOG descriptor combined with motion parameters is a very suitable car detector which reaches in record time a satisfactory recognition rate in dynamic outside area and bypasses several popular works without using sophisticated and expensive architectures such as GPU and FPGA. Also the performances of enhancement on road car detector are evaluated by the comparison of estimated cars positions with a ground truth manually extracted ones from a challenging database integrating a cars numbers and directions variations. The experimental study proves that estimated values are very close to the ground truth ones.