Int J Performability Eng ›› 2019, Vol. 15 ›› Issue (2): 688-699.doi: 10.23940/ijpe.19.02.p33.688699

Previous Articles     Next Articles

Vehicle Detection on Unmanned Aerial Vehicle Images based on Saliency Region Detection

Wenhui Lia, Feng Qua*(), and Peixun Liub   

  1. a College of Computer Science and Technology, Jilin University, Changchun,130021, China
    b Changchun Institute of Optics, Fine Mechanics and Physics; Chinese Academy of Sciences, Changchun, 130033, China
  • Revised on ; Accepted on
  • Contact: Qu Feng E-mail:qufeng_jlu@163.com

Abstract:

The target detection and tracking technology of the unmanned aerial vehicle (UAV) is an important research direction in the field of UAV aerial photography. In order to effectively and accurately detect vehicles in a UAV platform and in complicated road environments, the authors proposed a vehicle detection method based on saliency region detection. First, the saliency map of the target is calculated by using the salient region detection method based on the optimized frequency-turned. Next, segmentation methods based on Boolean Map and OTSU are combined to determine the region of interest of the vehicle target in the saliency map image. Finally, a series of vehicle apparent features-based methods based on geometry, symmetry, and horizontal edge wave are used to determine the vehicle and eliminate the interference of roadside objects accurately. Experimental tests carried out from different datasets show excellent performance in multi-vehicle detection in terms of accuracy in complex traffic situations and under different scales and angles of aerial images, realizing fast vehicle detection on the UAV platform.

Key words: unmanned aerial vehicle, free motion camera, vehicle detection, salient region detection