基于二次阈值的荔枝果与结果母枝图像识别  

Recognition on image of litchi fruits and their main fruit bearing based on twin-threshold method

作　　者：郭艾侠[1,2] 邹湘军[1] 邹海鑫[1] 

机构地区：[1]华南农业大学南方农业机械与装备关键技术教育部重点实验室,广东广州510642 [2]华南农业大学信息学院,广东广州510642

出　　处：《计算机工程与设计》2014年第2期557-561,共5页Computer Engineering and Design

基　　金：国家自然科学基金项目(31171457)

摘　　要：为解决荔枝收获机器人对采摘目标识别与定位的关键问题,给出了荔枝图像感兴趣区域的荔枝串、荔枝果与结果母枝的图像分类,进行了彩色荔枝图像基于YCbCr色彩空间的Cr通道图的灰度化处理,提出了基于二次阈值的图像分割方法分类识别荔枝各部位的策略。对强光、逆光和普通光照条件下所采集的90幅荔枝图像进行荔枝串、荔枝果与结果母枝分类识别的实验测试与分析,各部位的平均识别率分别为91.67%、91.67%和86.67%。实验结果表明了该方法对荔枝不同部位的识别有效、可行。To solve the key problems on recognition and positioning of picking object for litchi harvesting-robot, image sorts on litchi cluster, litchi fruits and their main fruit bearing branch in the region of interest of litchi image are given, and grayed color litchi image to Cr Single-channel image of YChCr color space is processed, and segmentation strategy based on the twin-threshold method to recognize all parts of litchi image is proposed. Finally, testing experiment of recognizing cluster, fruits and their main fruit hearing branch of litchi was carried out, with respective average ratio of 91.67~, 91.67~/oo and 86.67%, by taking 90 col- lected images influencing illumination in highlight, normal light and backlighting as test object. Results of both test on recogni- tion of all parts of litchi and data analysis attested that method proposed in the article is feasible and effective.

关 键 词：收获机器人 结果母枝 荔枝 阈值分割 YCBCR色彩空间 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]