Vibration and Reflection Reduction in Images Captured from Moving Objects Connected Through a Wireless Sensor Network

This research explores the use of a computational approach to stabilize image sequences captured by vision sensors mounted on unmanned aerial vehicles (UAV) for road traffic surveillance. The images captured are processed in real-time, and the information

  • PDF / 745,820 Bytes
  • 11 Pages / 439.37 x 666.142 pts Page_size
  • 77 Downloads / 217 Views

DOWNLOAD

REPORT


Abstract This research explores the use of a computational approach to stabilize image sequences captured by vision sensors mounted on unmanned aerial vehicles (UAV) for road traffic surveillance. The images captured are processed in real-time, and the information necessary to create a feedback to stabilize the flight and enhance the image is computed. The proposed approach is tested on Zigduino, a prototype vision sensor developed for wireless sensor networks used in quadcopter UAV. The results show that our approach can be applied to scenes that have moving objects and dynamic luminance. Although our focus is on UAV for road traffic surveillance, the D. Afolabi (B) · N. Zhang · H. N. Liang · K. L. Man · D. Liu · E. G. Lim · T. O. Ting Xi’an Jiaotong-Liverpool University, Suzhou, People’s Republic of China e-mail: [email protected] N. Zhang e-mail: [email protected] H. N. Liang e-mail: [email protected] K. L. Man e-mail: [email protected] D. Liu e-mail: [email protected] E. G. Lim e-mail: [email protected] T. O. Ting e-mail: [email protected] Y. Yang · L. Cheng Suzhou Institute of Nano-Tech and Nano-Bionics (SINANO), Chinese Academy of Sciences, Suzhou, People’s Republic of China e-mail: [email protected] L. Cheng e-mail: [email protected] G.-C. Yang et al. (eds.), Transactions on Engineering Technologies, Lecture Notes in Electrical Engineering 275, DOI: 10.1007/978-94-007-7684-5_6, © Springer Science+Business Media Dordrecht 2014

71

72

D. Afolabi et al.

results of this research have applications to a variety of moving objects which have mounted cameras to capture images and transmit these wirelessly. Keywords Glare reduction · Hough transform · LooCI · Traffic surveillance Vibration reduction · Wireless vision sensor network · Zigduino

·

1 Introduction In designing a wireless sensor network (WSN) of unmanned aerial vehicles for traffic surveillance, a number of important issues should be addressed in order to have a robust system that is capable of capturing high quality images regardless of environmental conditions (e.g., wind speed, sunlight intensity / luminosity). Miniature quadcopter drones have been commonly used for surveillance. They can offer added functions over traditional means of stationary mounted cameras such as enabling continuous tracking of multiple vehicles on the road, aiding coordination among multiple drones within a region of interest, and detecting drunk driving and other road traffic anomalies [1]. Due to limited resources in most WSN, such as power, computational capabilities, and transmission bandwidth, the framework needs to be designed based on low-power, low-cost, and commercial off-the-shelf products. To meet these considerations, therefore, the wireless vision sensor network (WVSN) in this research has been developed using the Zigduino board which is an Arduino variant opensource electronic platform that includes an ATmega128RFA1 microcontroller with an in-built 2.4 Ghz IEEE 802.15.4 transceiver. The Zigduino devic