Abstract. The work presents a gray-scale image recognition system developed to process images from a digital video camera. The aim of the system is to identify the position of the robots and the ball in a robot soccer competition. The image recognition strategy is based on histogram analysis technique. Initially, the image is processed to find object edges containing the same gray-scale characteristics of the target object. Subsequently, the system verifies the similarities between the obtained object and the target. The experimental results are promising and show that this technique can be successfully applied to real time robot recognition in robot soccer competitions. The basic principle for localizing the elements (players and ball) on the soccer field is based on histogram analysis technique. This process has three main phases. The first one is to define the border of any object in the image that has the same gray level of the target object. In the second phase, the object border is analyzed (considering size and the proportion of its dimensions) to differentiate between a robot and the ball, or any other object in the field (noise, shadows, etc.). Finally, in the third phase, the robot marker is analyzed for its shape, size and gray-scale level to identify which robot it is and to which team it belongs.
[ project home page | contact authors ]
 |
The Advanced Studies Group on Robotics of the Mathematical and Computer Sciences Institute at USP São Carlos has the objective of improving the technical abilities of the under graduation students in some areas such as Electrical Engineering, Aeronautical Engineering, Mecatronics, and Computer Sciences, for the development of autonomous robotic systems that perform in robot soccer competitions. |