基于偏好免疫网络的油茶果采摘机器人图像识别算法

摘要

With the demanding of morphological features recognition in picking robot machine-vision system, preference artificial immune net ( aiNet) was used as the image recognition algorithm, mainly modified the structure of the algorithm to promote the accuracy rate. The algorithm was modified according to picking environment and real-time require. The simulation proved the clustering accuracy of preference aiNet reached to 81. 67% in the sunny day and 87. 69% in the cloudy day. The modified algorithm has certain meaning in the next research of picking robot.%针对油茶果采摘机器人视觉识别中外界物体的形态学特性要求,采用了偏好人工免疫网络算法作为机器视觉的图像识别算法,并根据采摘环境及采摘对象的特点对算法结构进行了改进,增强了算法的识别率.仿真实验表明,采用偏好人工免疫网络算法对油茶果的识别率在晴天时达到了81.67％,阴天时达到了87.69％,满足采摘识别率的要求.