A real-time moving object detection using wavelet-based neural network
JongBae Kim · 2009
This paper presents a real-time moving object detection method using the wavelet-based neural networks for the pre-crash safety system of a vehicle. The proposed method uses stationary infrared cameras to sense vehicles and obstacles on the road ahead, and the CarPC to determine whether or not a collision is based on the speed, position, and traveling route of the object. In this method, the stationary infrared camera offers stereo image, and possesses good results even in inclement weather such as night or rain. The procedure toward complete object detection consists of following steps: pyramid representation, image segmentation, local matching, object recognition, and disparity estimation. The proposed method can be useful for applying the intelligent vehicle system.