Difference between revisions of "VEDO"
| (One intermediate revision by one other user not shown) | |||
| Line 8: | Line 8: | ||
|start=2011/09/10 | |start=2011/09/10 | ||
|end=2011/12/22 | |end=2011/12/22 | ||
| − | |status= | + | |status=Closed |
|level=Ms | |level=Ms | ||
|type=Thesis | |type=Thesis | ||
| Line 23: | Line 23: | ||
The detection is possible due to the use of a model containing the main features of the remote control. This model is sought in every considered frame of the video stream and the entire process is iterative. | The detection is possible due to the use of a model containing the main features of the remote control. This model is sought in every considered frame of the video stream and the entire process is iterative. | ||
An output stream is produced and it is composed by every analyzed input frame. | An output stream is produced and it is composed by every analyzed input frame. | ||
| − | When the object is detected, a rectangular is | + | When the object is detected, a rectangular is showed around it, so the user can have a proof of the correct detection. |
| − | This whole procedure is performed in order to get | + | This whole procedure is performed in order to get fast and robust results, while respecting the constraints of time and trying to exploit, in the best way, the hardware on board of Parrot. |
The entire project is coded in C language and using the OpenCV 2.3.1 libraries under Linux. | The entire project is coded in C language and using the OpenCV 2.3.1 libraries under Linux. | ||
Latest revision as of 22:14, 8 January 2012
VEDO
| |
| Coordinator: | AndreaBonarini (andrea.bonarini@polimi.it) |
| Tutor: | AndreaBonarini (andrea.bonarini@polimi.it) |
| Collaborator: | |
| Students: | AlbertoBottinelli (bozzolino1@vodafone.it), RoccoDato (roiko@hotmail.it) |
| Research Area: | Robotics |
| Research Topic: | Robot development |
| Start: | 2011/09/10 |
| End: | 2011/12/22 |
| Status: | Closed |
| Level: | Ms |
| Type: | Thesis |
VEDO is an Enhanced Detector of Objects
Aim of the project is to develop a system based on quadrotor to find an object in an indoor environment. The image analysis system has been implemented off-board, as an extension of the Viola-Jones algorithm.
In particular, the main purpose of this project is related to the detection of an ordinary object by using the frontal camera onboard a flying Parrot AR.Drone quadcopter. The system must be able to detect the object even if it's partially hidden. The scenario considered during this project development expects to find a remote control located in a room and positioned in an unknown location. By using of a controller, the user drives the quadcopter while an application we developed allows to find the remote control. This application analyzes the video stream provided by the frontal camera of the Parrot. The detection is possible due to the use of a model containing the main features of the remote control. This model is sought in every considered frame of the video stream and the entire process is iterative. An output stream is produced and it is composed by every analyzed input frame. When the object is detected, a rectangular is showed around it, so the user can have a proof of the correct detection. This whole procedure is performed in order to get fast and robust results, while respecting the constraints of time and trying to exploit, in the best way, the hardware on board of Parrot. The entire project is coded in C language and using the OpenCV 2.3.1 libraries under Linux.
Demo movie: an application to the retrieval of a remote control, partially covered.
