Research On Vision Of Humanoid Soccor Robot Based On ARM

In the field of humanoid soccer robot, the target recognition and tracking has becomeone of the most important topics. How to identify the target in the image is a key factor forsoccer robot. Based on analysis of the present situation of the domestic and foreignresearch, this paper comprehensively uses many aspects of theories knowledge and doesstudies on the approach for soccer target recognition and target tracking.Firstly, a design of hardware of the humanoid soccer robot is given. This paper putsforward a design scheme of double processors. The processor of ARM is used to captureimage and process image. The processor of51single-chips is used as the servo controller.Between ARM and51single-chips use the serial to communicate with each other.Secondly, a software platform of humanoid soccer robot is built. The softwareplatform includes the embedded operating system of Linux and the image soft package ofOpenCV. The platform capture image by using the protocol of V4L2. The platformprocess image by using OpenCV.Thirdly, an improved color target recognition algorithm is given. The doublethreshold algorithm is combined with seed growing algorithm to produce the improvedcolor target recognition algorithm. The double threshold algorithm includes the fixedthreshold and the dynamic threshold. The fixed threshold is obtained by processing theimage of target sample. According to the change of illumination, the dynamic threshold isupdated. The intersection of the fixed threshold and dynamic threshold is the doublethreshold. The experiments are carried out in the humanoid soccer robot, verifying theaccuracy of the new algorithm, and the analysis of the experimental results is made.Fourthly, as the humanoid soccer robot tracking target is single color ball, animproved CamShift algorithm is proposed. The algorithm obtains the target initial positionand size by using recognition algorithm. According to the long axis and short axis of target,the new algorithm calculates the circular degree of target, and judge whether the trackingtarget is correct. The experimental of is made by using the humanoid soccer robot,verifying the feasibility and stability of the new algorithm, and the analysis of the experimental results is made.