摘要
针对传统瓷砖人工检测工艺和判据单一化问题,在MATLAB开发性工作环境中,鉴于神经网络所具有的自组织、自学习和分布式存储信息的特点,提出了一种基于神经网络综合处理颜色特征和形状特征进行瓷砖图像检测的方法,利用Matlab神经网络工具箱建立特征参数检测模型,检测图像中的缺陷。首先,根据分类要求选定合适的神经网络结构,依据产品对象选择最基本的特征参数,并通过试验总结出最佳的算法设计。最后,运用该方法对sonydscp73型号照相机所采集的产品图像进行实验。实验证明,神经网络为瓷砖检测提供了一个很好的工具。
In the MATLAB growth oriented working conditions, aiming at the problem of manual examination craft in traditional tries detection and criterion singleness,this article put forward a method of tiles detection which synthetically processed color features and shape features of tile＇ s image base on neural network because of its characteristic of self-organizat ion selfstudy and dist ribut ion sto rage information. Matlab NN Too 1 is used to build the characteristic parameter model which detects the flaw of input image. First, According to classified request to choose appropriate neural network structure, and chooses the most basic characteristic parameter based on the product object, and summarizes the best algorithm design through the experiment Finally, the product ima-ge utilizes that method which gathers to the sonydscp73 model camera to carry on the experiment. The result shows neural network provides a good tool to the tile＇ s detection.
出处
《中国陶瓷》
CAS
CSCD
北大核心
2008年第12期52-54,共3页
China Ceramics
关键词
MATLAB
瓷砖
神经网络
Matrix Laboratory
ceramic tile
neural network
