Vehicle Detection From Aerial Images With An New Algorithm

Zezhong Zheng,Xiaoting Wang,Wenfeng Zheng,Guoqing Zhou
2010-01-01
Abstract:Vehicle detection from high resolution aerial images has been studied for many years. However, a robust and efficient vehicle detection is still challenging. In this paper, a novel and robust method for automatic vehicle detection from aerial images was presented. In this method, a GIS road vector map is used to constrain a vehicle detection system to road networks, edge detection and morphological preprocessing method are used to identify candidate vehicle pixels. Different types of vehicle templates are chosen to adaptively solve their relative regular rotation angle to detect the similar vehicles by their correlation coefficient with the same size of the windows of the templates. Experiment is conducted using 0.15 meter resolution aerial images, the result demonstrated that the new method has a excellent detection performance.
What problem does this paper attempt to address?