基于双目视觉的棉花三维重构技术

摘要

Aiming at the vision system of cotton picking robot , it studies the three-dimensional reconstruction of cotton based on binocular stereo vision method .Using the CCD camera imaging model and Zhang Zhengyou plane calibration method , it calibrates the camera .Combining the block average edge detection and noise classification method with median filtering and histogram PCNN (Pulse Coupled Neutral Net -works) segmentation, it realizes the pretreatment of cotton image.Using SIFT(scale invariant feature transform) algorithm, it extracts the feature points around the camera image and realizes stereo matching .The results of 3 D reconstruction of cotton feature map are obtained in the end .%针对采棉机器人中的视觉系统,以双目立体视觉方法对棉花三维重构技术进行研究.通过CCD摄像头光学成像模型和张正友标定法对摄像机标定,采用分块平均边缘检测和噪声分类、中值滤波、直方图的PCNN分割方法对棉花图像进行预处理,运用SIFT算法提取左右摄像机图像特征点并进行立体匹配,最后得到了棉花特征点的三维重构图.