【摘要】 针对车用换热器表面缺陷检测问题,提出一种利用傅立叶图像重建提取方向纹理表面缺陷的算法。该算法采用局部二值模式(LBP)描述去除光照影响并对LBP图像进行快速傅立叶变换,通过频率域滤波和图像重建有效去除方向纹理的影响,由二值化及团块分析最终提取换热器表面缺陷区域。实验结果证明,本文算法可以快速、有效地检测出换热器表面缺陷。更多还原

【Abstract】 To find the deficiency on the surface of a vehicle heat exchanger,this paper has brought out a defect detection algorithm for directional texture surface based on image Fourier reconstruction. The algorithm first performs the local binary pattern( LBP) transformation on original image to alleviate uneven lighting condition. Then,the fast Fourier transformation is used to transform the LBP image into spectrum domain. Through spectrum filtering by which all high-frequent coefficients and coefficients with high energy are suppressed,directional texture are eliminated on the Fourier reconstructed image. Thus,a general binarization process followed by a blob analysis can be employed on the result homogenous image to finally extract defect areas. Experimental results have demonstrated the effectiveness and swiftness of the proposed algorithm.更多还原