基于相位一致性的低对比度纸病识别算法研究

Identification Algorithm of Low Contrast Paper Defects Based on Phase Congruency

针对机器视觉检测低对比度纸病,存在常规的阈值分割会引起低对比度纸病信息丢失以及边缘检测存在鲁棒性差的问题,本课题提出了一种基于相位一致性算法识别低对比度纸病的方法,并与常规的阈值分割以及边缘检测中具有代表性的canny算子进行了对比分析。结果表明,当识别低对比度纸病时,本课题提出的方法不仅保留的有用信息较常规阈值分割的多,而且鲁棒性较canny算子的边缘检测好。

According to the phenomenon of losing some important information in conventional threshold segmentation and poor robustness in edge detection when detecting low contrast paper defects based on machine vision,this paper proposed a method to identify low contrast paper defects based on the idea of phase congruency,and verified the feasibility.In this paper,not only the feature extraction method of phase congruency of low contrast paper image was given,but also the generation method of adaptive threshold used in segmentation was given,and compared with the conventional threshold segmentation and the representative canny algorithm in edge detection in experiment.The experimental results showed that when detecting low contrast paper defects,the proposed method preserved more useful information than the conventional threshold segmentation,and the robustness was better than canny edge detection.