Motion Detection of Small Objects
MetadataShow full item record
The paper describes a problem of detection small moving objects. A size of these objects is only one or few pixels. Relevant mathematical apparatus which are usually used for processing of dynamic pictures are included. There was proposed experimental module for standalone locator for real time operation. This module is consisted from two main parts. First part is used for imaging of scene, especially small moving objects. We use four camera systems. The Second part comprises a development board based on Raspberry PI 2. We use GNU/Linux system on the board. There were also tested methods useful for the essence of this project. On behalf of the tests was developed an algorithm to detect small particle in video sequence. C and python language routines were build including essence algorithms. This algorithm consists of: images are captured one by one from all 4 cameras 640 £ 480; RGB to Gray transformation; For each image are all shapes detected; unnecessary shapes are removed; Images are filtered with additional filters; Noise free images are then compared for movement detection. Images with movement are stored on SD card. Additional information: Devices captures and stores all flying objects. Data are accessible through LAN port. We use OpenCv for image analysis. Device is used for sky movement detection of small objects, birds. After detection, it can create acoustic signal to scare them away. This minimizes bird impact in Agriculture and it prevents bird-plane collisions on airports. This paper describes in more detail not only the entire detection algorithm, but also the results of the test algorithm.
Document typePeer reviewed
Document versionxmlui.vut.verze.Publishers's version
SourceProceedings of the 22nd Conference STUDENT EEICT 2016. s. 92-94. ISBN 978-80-214-5350-0
- Student EEICT 2016