摘要
目前新疆南疆地区红枣分级大多采用原始的人工分级方法,分级效率低,漏检率高,劳动强度大,影响红枣品质指标的评价。针对这一现状,提出采用机器视觉红枣纹理分级方法,提高红枣采后分级处理能力,加快红枣产业化发展。试验选用300个红枣图片作为学习训练,60个红枣图片作为测试,应用BP神经网络进行红枣纹理分级。结果表明:BP人工神经网络分级与人工分级的一致度达到93.33%。
At present, it is an original manual grading methods of jujnbe classification adopted mostly in south Xiniiang area, low classification efficiency, high omission rate, high labor intensity, effecting jujube quality index evaluation. In view of the present situation, the machine vision texture classification method is proposed, which improves the processing ability of iuiube mining, and accelerates the development of jujube industry. 300 pictures of jujubes were assigned to be trained, 60 to be tested. The BP neural network is adopted to classify the juiubes based on the texture feature. The result shows that, consistency of BP artificial neural network and artificial classification reached 93.33%.
出处
《中国农机化学报》
2016年第3期201-204,共4页
Journal of Chinese Agricultural Mechanization
基金
新疆库尔勒市重点科技项目(2012-8)
关键词
机器视觉
人工神经网络
纹理分级
红枣
machine vision
BP neural network
texture classification
jujube
