By Topic

Bird Objects Detection and Tracking on the Wild Field Circumstance

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$33 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

3 Author(s)
Yan Li ; Int. Inst. for Earth Syst. Sci., Nanjing Univ., Nanjing, China ; Yongyi Yin ; Xu, Bo

Poyang lake of Jiangxi province, China, is the biggest wetland of China. It is one of the largest bases of the migrant birds especially in winter. The nature ecosystem of the region has relationship with the birds' activities, and vice versa. Research of the birds is one of the important ways for knowing the ecosystem. We attempt to set up a system of surveillance and estimation of the bird types and amounts from video. The background of the landscape is the wild field, mainly including lake, grass, and sky. The video camera is the handy one. Because of the wild birds such as the Cygnuspsilas nature, they are likely settle in a large amount in the lake bay and are far away from the observers. The amount may be over 10,000 that the view of field has to pan slowly in the horizontal to capture all of them. So we have to face up with the problem of the moving background together with the small and crowd objects. In this paper we propose a method based on adaptive filtering and morphological algorithm for moving object detection. It aims at relevant low complex of the computation and also robust for most scenes. Firstly the camera motion is eliminated using motion compensation. The predict frame is estimated from the last frame, and difference image is obtained by filtering of the predicted frame and the current frame. The pixels of the moving objects show high values. The distribution of the difference is modeled as a Gaussian function and a segmentation threshold is calculated adaptively according to the distribution function. Then the segmented moving pixels are detected and dilated to construct approximate parts of objects. At last, motion directions and speeds of the moving parts are computed by matching them of the sequent frames and be used to cluster the parts into objects. The experiment shows the efficiency of this method.

Published in:

Computer Science and Information Engineering, 2009 WRI World Congress on  (Volume:6 )

Date of Conference:

March 31 2009-April 2 2009