摘要
针对带钢表面的划伤、黑斑、翘皮、辊印、褶皱和压印6种典型缺陷,提取样本图像的灰度、纹理和几何形状特征等20维特征向量;给出粗糙集理论的关键技术,基于粗糙集理论构造带钢表面缺陷图像识别的决策表,对决策表进行属性约简,并直接从训练样本图像中导出决策规则;应用所获取的规则对带钢表面缺陷测试样本图像进行分类,并同BP算法进行对比,验证了基于粗糙集理论的分类识别算法的有效性。
The 20dimensional feature vectors of intensity, texture and geometry characteristics for six kinds of steel strip surface typical defects images are extracted. The key technology of Rough Set theory is described. The decision table of the steel strip surface images recognition is created, the reduction for decision table is carried out, and the decision rules are obtained from the training sample images directly. The test samples of the steel surface defect images have been classified with application of decision rules, and then compare with the BP neural network algorithm. The recognition and classification of steel strip surface typical defects images based on rough set theory is effective.
出处
《中国图象图形学报》
CSCD
北大核心
2011年第7期1213-1218,共6页
Journal of Image and Graphics
基金
高等学校博士学科点专项科研基金项目(20104219110001)
武汉市科技攻关项目(200910321100)
武汉科技大学青年科技骨干培育计划项目(2009xz24)
关键词
粗糙集理论
带钢表面缺陷
识别
分类
rough set theory
steel strip surface defects
recognition
classification
