摘要
如何对采集到的草莓图像进行分割和如何选取评定草莓等级的特征参数是草莓自动分拣系统的2个重要环节。该研究利用草莓R、G、B通道明显的像素差值来分割目标和背景,并且选取草莓的形状特征和成熟度作为草莓评级的特征参数,综合运用机器视觉、神经网络等理论方法,通过实验数据统计,建立极坐标下草莓外形轮廓特征参数及颜色空间下成熟度特征参数的提取方法,以人工神经网络为识别模型,实现对草莓的自动分类。实验结果表明,该方法对草莓的自动分级结果与人工分级结果相比较,准确率达到90%,具有实际的可行性。
How to partition the collected images of strawberry and how to select the parameter which can be assessed the level of strawberry are the two important parts of automatic sorting system of strawberry. The object was segmented from background by using clear pixel value difference of R, G, B strawberry＇ s channels. And the shape feature and the degree of maturity was selected as the characteristic parameters of strawberry rating. The machine vision, neural network theory and method were integrated. Through the experimental data stalistical method, we set up the method that how to extract the characteristic parameters of the strawberry＇ s color space maturity and contour under the polar coordinates. After testing, by comparing with the manual classification results, the accuracy rate of rating strawberry which be made by the automatic sorting system reached 90%, which proved that the method has practical feasibility.
出处
《安徽农业科学》
CAS
2015年第21期370-373,共4页
Journal of Anhui Agricultural Sciences
基金
大学生创新创业训练项目(201210674030)
关键词
草莓
图像采集
图像分割
特征提取
Strawberry
Image acquisition
linage segmentation
Feature extraction
