自然场景下花椒果实的识别  

Identification of Chinese prickly ash under the natural scenes

作　　者：万芳新[1] 白明昌[1] 贺志洋 黄晓鹏[1] 

机构地区：[1]甘肃农业大学工学院,兰州市730070

出　　处：《中国农机化学报》2016年第10期115-119,共5页Journal of Chinese Agricultural Mechanization

基　　金：甘肃省自然科学基金资助项目(145RJZA041)--花椒采摘机械损伤机理及气流无损采收技术研究;甘肃省高校科研资助项目(2014A-063)--气吸式花椒采收机理研究;陇原青年创新人才扶持计划项目--气吸式花椒采收机理及配套机具的研制

摘　　要：提出一种自然场景下应用于花椒采摘机器人视觉系统的果实识别算法。该方法采用红色和绿色图像差异(R—G)来进行图像分割,使用优化腐蚀滤除噪点和隔离带法除去非目标花椒串果实,再求取花椒果实串的形心和半径。试验结果表明,该算法能有效的识别花椒果实,确定果实串的中心位置误差为10mm,识别半径误差与花椒果实串最小半径比0.05。Recognition algorithm of Chinese prickly ash fruit under nature scenes for harvesting robot was presented.The difference between red and green images（R-G）was used to segment the image,the optimized corrosion filter was used to filter noise,the blank tape as a symbol was used to filter non-objective fruit,and then the center and the radius of Chinese prickly ash train were calculated.The Chinese prickly ash fruit can be effectively identified by using this algorithm,the error of center position is 10 mm,and the ratio of identified radius error and the minimum fruit strain radius is 0.05.

关 键 词：花椒 自然场景 果实识别 

分 类 号：S817.12[农业科学—畜牧学]