自适应优化Gabor滤波器的带钢表面缺陷分类

Self-adaptively optimized Gabor filter for classification of strip surface defects

带钢表面缺陷纹理存在复杂性、多样性,导致对带钢表面纹理缺陷进行分类十分困难,为此,提出一种基于鲸鱼群算法的自适应优化Gabor滤波器。首先,利用各项异性扩散滤波抑制缺陷图片之中的伪边缘,再以不同种类缺陷特征的类间差最大作为目标函数,Gabor滤波器的参数为优化变量,采用鲸鱼群算法对Gabor参数进行寻优;然后,将所得到的Gabor特征进行融合;最后,导入分类器之中进行分类。实验结果表明,该方法具有较好的区分性和鲁棒性,针对常见的带钢表面缺陷,如冲孔、污渍、刮边、黑氧化条、结疤等最终的分类精度能达到97.5%。

The surface texture of the strip is complicated,which makes it difficult to classify the surface texture defects of the strip.For this reason,an adaptively optimized Gabor filter based on the whale swarm algorithm is proposed.The anisotropic diffusion filtering is used to suppress the false edges in the defect pictures,and then the maximum difference between the different types of defect features is taken as the objective function.As the parameters of the Gabor filter are optimization variables,the Gabor parameters are optimized by the whale swarm algorithm.The obtained Gabor features are merged and then imported into the classifier for classification.The experimental results show that the method has good discriminability and robustness.For the common strip surface defects named punching,stains,scraping,black oxide strips and crusting,the final classification accuracy can reach 97.5%.