Color Based Object Segmentation on Wheeled Goalkeeper Robot

摘要

The ball detection system is a basic ability that must be possessed by soccer player robots, including goalkeeper robots. With the ball detection system, the robot can find out the position of the ball on the field. The position of the ball relative to the goalkeeping robot and the direction of its motion can affect the decisions and movements that will be made by the robot. In other words, the ball detection system has an important role for the goalkeeper robot in blocking the ball that threatens the goal. The problem that is often encountered with this detection system is, the system will detect all objects of the same color in the image even outside the field. This causes the robot to process the wrong information. To overcome this error, an elimination process is carried out on the area outside the field in the input image. This paper presents an object segmentation and detection system. To get the position of the object, it is done by calculating the moment of the contour. After doing some experiments, the result obtained that system successfully eliminates some areas outside the field from the image. This system also only detects the ball and robot in the field, with an average of processed frames per second (fps) is 31 fps.