基于PCA模式和颜色特征的钢轨表面缺陷视觉显著性检测

摘要

针对现有机器视觉缺陷检测方法存在缺陷显著性不明显、鲁棒性较弱等问题,提出一种主成分分析(PCA)模式和颜色特征相融合的钢轨表面缺陷视觉显著性检测方法.为解决传统模式特征计算效率低的问题,对PCA模式特征进行改进.利用缺陷图块的颜色特征与锈迹、斑痕等普通图块的差别较大这一特点,对钢轨表面缺陷进行显著性检测.试验结果表明,该方法可以处理不同形状的轧疤、轧痕类缺陷,缺陷显著性效果较好,能准确地显示缺陷形状、位置等信息.%In view of the existing problems of unobvious defect saliency,weak robustness, etc. ,the visual saliency detection method based on PCA mode and color features for rail surface defects is proposed. To overcome the disadvantage of low efficiency in traditional mode feature calculation, the improvement is conducted by using PCA mode feature. The color features of the defect image blocks are quite different from the ordinary image blocks of the rust and scar, so the saliency of the surface defects can be detected. The test results indicate that this method can be applied to deal with the defects of rolling scar or rolling marks with different shapes, and offers better defect saliency; in addition, it can display more information about the shapes and locations of the defects.