Edge-based Forward Vehicle Detection Method for Complex Scenes

A video-based vehicle detection method must be capable of continual operation under various environmental and illumination conditions. Therefore, identifying features that can adapt to various conditions is crucial. This paper proposes a robust vehicle detection method for identifying horizontal edges by using a Sobel filter to achieve low computational complexity. Based on the orientation of the gradient, the feature of a vehicle can be extracted accurately and quickly. Additionally, symmetrical features are critical features, and a histogram of oriented gradients was employed to reduce the detection error rate. The experimental results indicate that the proposed method can efficiently and effectively detect forward vehicle in various scenes.