| Summary: | 碩士 === 國立臺北科技大學 === 自動化科技研究所 === 99 === In this paper, fuzzy-based novel real-time color detection method and robot soccer competition vision system is explored for robot soccer competition. At first, according to the rules of competition, we use a camera which installed to the top of the field to capture the dynamic environment images, and transfer the data to the computer for recognizing the image and producing the decision to control soccer humanoid by wireless transmission. In the game, the locations of the ball and the humanoid are identified through the image system in a strong dynamic environment; therefore, it is very important to identify objects real-timely and accurately for an image system. Thus, we implement an interface including field correction, image identification and decision analysis which can identify the image real-timely and accurately. In the game, each identify object has the unique color, so the first step on the object recognition is the color identification. For the real world, the color clusters of object with glossy materials will deviate from the linear model and will cause the difficulty of identification in the different brightness. Therefore, in this paper, we propose a fuzzy-based novel real-time color detection method to improve this problem. Different from existing methods, the proposed method can improve in the robotics soccer image system which provides less execution-time and higher success rate to identify object. Finally, the experiment results are illustrated to demonstrate the proposed scheme can improve the efficiency in the robot soccer competition vision system.
|
|---|
