摘要
基于人工分拣的墙地砖质量检测环节不仅造成人力资源的浪费,更无法保证质量检测的准确度,影响了墙地砖产品的档次提高。为了节省成本,进一步提高墙地砖的生产效率,本文利用颜色通道下的共生矩阵特征作为图像视觉特征,并充分利用图像的纹理信息和颜色信息,训练出一个适用于墙地砖缺陷分类的BP神经网络。通过实验结果的数据分析,基于BP神经网络的墙地砖缺陷检测技术能够对多种尺寸规格、颜色、图案的墙地砖得到较好的检测结果。
The quality testing session of tiles based on manual sorting not only causes waste of human resources, but cannot guarantee the quality of detection accuracy, affecting the improved quality of wall and floor tiles. In order to save costs and further improve the production efficiency of wall and floor tiles, in this paper, the features of co-occurrence matrix under color channels is taken as image visual features, and by taking advantage of image texture and color information, a BP neural network that applies to defect classification of wall and floor tiles is trained. Through data analysis of experimental results, the wall and floor tile detection technology based on BP neural network can get a better test results for a variety of sizes, colors, patterns of wall and floor tiles.
出处
《微型机与应用》
2014年第23期81-83,共3页
Microcomputer & Its Applications
关键词
颜色通道
共生矩阵特征
墙地砖缺陷
BP神经网络
color channels
the features of co-occurrence matrix
the defects of wall and floor tile
BP neural network
