蛇形缠绕机理及其在视觉定位中的应用

The Mechanism of Snake-Enlacing and Its Application to Visual Positioning

摘要: 为解决自动灌装系统的快速视觉定位问题，本文通过对蛇类缠绕行为中的力学分析和过程模拟，提出了一种孔心仿生定位方法。该方法先对灌装桶图像进行快速二值化处理，并针对噪声和灌注孔形成的黑色像素，获取其重心位置；随后，模拟蛇在缠绕猎物时的“挤压过程”，缩小图像处理范围；并通过蛇的“姿态调整”行为，重新寻找图像重心，直至图像重心与灌注孔心完全重合。对比实验表明：本文方法计算量小，精度高，抗干扰能力强，处理速度是Hough变换方法的20余倍。

Abstract: To solve the problem of rapid visual positioning in automatic filling system, a bionic positioning method for the filling center is proposed based on the force analysis and process simulation of snake-enlacing. In the proposed method, the image of filling barrel is firstly transferred into a binary one by means of a fast algorithm. Then the image is narrowed by simulating the way that snakes crimp their preys after the centroid of black pixels caused by the filling hole and noise is calculated. And the centroid of the narrowed image is recursively calculated inspired by the behavior of gesture adjustment of snake until the filling center is in accord with the resulting centroid. The comparison experiment shows that the proposed method is with less computational consuming, higher precision, robust against the noise, and high calculation speed more than 20 times than the Hough Transform method.