摘要
提出了一种用于砂纸、铸件和许多工业材料中的随机纹理表面缺陷的自动检测的全局方法。该方法不依赖于纹理的局部特征,它应用傅里叶变换进行全局图像恢复。应用逆傅里叶变换去除任何统计纹理中的周期性、重复性结构。在恢复图像中,原图像中的同质区域灰度近似一致,而缺陷区域被明显地保留下来了。对不同实际统计纹理的实验结果表明了该方法的有效性。
This paper brought forth an overall scheme of automatic detecting, which was applied to the surface of stochastic texture, including sand paper, casting and many industrial materials. Based on the overall image reconstruction method of Fourier transform, the scheme is independent from the partial feature of texture, with its effectiveness testified by outcomes of a number of tests on various statistical textures.
出处
《计算机应用研究》
CSCD
北大核心
2007年第8期193-194,197,共3页
Application Research of Computers
基金
广东省自然科学基金资助项目(7007362)
国家自然科学基金资助项目(50475044)
广州市黄埔区科技计划资助项目(0713)
关键词
表面检测
缺陷检测
统计纹理
傅里叶变换
图像重构
surface-detecting
defect-detecting
statistical texture
Fourier transform
image reconstruction
