基于OHTA颜色空间的瓜果轮廓提取方法  

OHTA Color Space Based Method for Fruit Contour Detection

作　　者：郭峰[1] 曹其新[1] 谢国俊[1] 周金良[1] 

机构地区：[1]上海交通大学机器人研究所

出　　处：《农业机械学报》2005年第11期113-116,共4页Transactions of the Chinese Society for Agricultural Machinery

基　　金：上海市科技兴农重点攻关项目(项目编号:农科攻字(2001)第21号)

摘　　要：提出了一种对彩色水果图像进行果实轮廓提取的算法。首先将图像转换为OHTA颜色空间中的灰度图像,用阈值法实现了图像分割;然后用BLOB算法滤去图像中的噪声;最后用插值的方法得到平滑的轮廓曲线。利用开发的智能瓜果精选分级试验样机,按照大小对温室西红柿进行分类,对算法的精度和实时性进行检验,结果表明算法的精度可达98%,分选速度达到3个/s,达到了实用化的目标。An algorithm to realize fruit contour detection in color image is presented in this paper. At first, a color image is transformed into a grey-level image in OHTA color space, and segmented by constant threshold value; then the BLOB algorithm is utilized to erase noises in the image; a spline interpolation based algorithm is adopted to get smooth fruit contour at last. To test the speed and accuracy of algorithm, 50 green-house tomatoes were classified according to size in the exploited intelligent fruit sorting experimental prototype. The experimental result shows that the accuracy of the algorithm is higher than 98%, its calculation speed is 3 samples per second, which satisfies practical requirements.

关 键 词：水果分级 机器视觉 轮廓提取 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]