自然环境下山核桃果实的图像识别提取方法  

The Method for Recognition and Extraction of Pecan Fruit in Natural Environment

作　　者：章云[1,2] 周素茵[1,2] 周竹[1,2] 

机构地区：[1]浙江农林大学信息工程学院,浙江临安311300 [2]浙江省林业智能监测与信息技术研究重点实验室,浙江临安311300

出　　处：《湖北农业科学》2016年第21期5643-5645,共3页Hubei Agricultural Sciences

基　　金：浙江省教育厅一般项目(Y201328071)

摘　　要：研究了自然环境下成熟山核桃果实的机器视觉图像识别方法。通过分析顺光、逆光、遮挡等环境下拍摄的山核桃成熟待采时的果实图像,确定了2R-G-B色差模型,用最佳阈值函数进行图像分割,经形态学滤波后采用Canny算子进行边缘检测,提取出复杂背景下山核桃果实的轮廓。试验结果表明,该方法对山核桃果实的有效识别率达86.7%,可为山核桃采摘机器人的研发提供技术参数。The image recognition method based on machine vision for mature pecan fruit in natural environment was researched.When the pecan fruit is mature,the dprofile of pecan fruit under complex background was extracted through the analysis on the images of fruit trees in direct light,backlight,shelter and other environment by using the 2R-G-B color model,adopting optimum threshold function to achieve image segmentation and selecting Canny operator for edge detection after morphological filtering. The experiment results showed that the effective recognition accuracy of the image recognition method for pecan fruit reached 86.7%,which can provide technical parameters for the research of the pecan fruit picking robot.

关 键 词：成熟山核桃果 机器视觉 图像识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]