柑橘果实信息智能识别及定位方法研究

On Information Intelligent Identification and Location Methods of Citrus

利用智能机器人进行柑橘采摘是提高采摘效率和节约劳动力资源的关键,因此果实信息智能识别及定位方法研究已成为目前研究的热点.目前主要的研究方法有:阈值分割算法、人工蜂群算法、色差分割算法、聚类算法及相关算法的组合算法.这些算法大多运算复杂,存储空间大,识别的精度低,尤其对复杂背景及多果重叠识别效果较低.以柑橘果实图像2R-G-B色差分量作为信息源,采取小波多尺度变换分割果实信息,同时使用MATLAB软件进行仿真,对于复杂背景及多果重叠情况下柑橘果实的识别及定位取得了较好的仿真结果.

The use of intelligent robot improves the efficiency and quality of fruit picking and saves labor resources,and the fruit information intelligent identification and positioning method research has become a current research hotspot.The main research methods are threshold segmentation algorithm,artificial colony algorithm,color segmentation algorithm,clustering algorithm and related combination algorithm.Most of them have problems such as complex in operation,large in storage space and low in recognition accuracy,however.Taking citrus image 2R-G-B chromatic component as an information generator,the paper uses multi-scale wavelet alternation to segment fruit information,makes simulation with MATLAB software for complex background and fruit overlaying situation of citrus identification and localization and has achieved very good results.