基于水平集的不规则透明表面缺陷检测设计

Design for Irregular Transparent Surface Defects Detection Based on Level Set

针对图像采集中,缺陷的灰度值和背景几乎不可区分,导致图像的低对比度的问题,目前,常规的表面缺陷检测图像预处理中常用阈值化和边缘检测技术,易出现漏检现象,即不适用于检测不规则表面边界处的细微瑕疵。为解决这一问题,从低对比度透明材料图像的非高斯统计特性出发,研究提出利用水平集智能分割具有复杂形状的目标对象,并利用小波变换进行缺陷特征提取的高效的智能缺陷检测方法。测试表明,所提出的基于水平集和小波变换的不规则表面缺陷检测系统的有效性,克服了检测区域边缘处的漏检缺陷问题。

For image acquisition,the grey value and background of defects are almost indistinguishable which lead to low contrast of the image.At present,thresholding and edge detection techniques are commonly used in the conventional image preprocessing of the surface defect detection,easy to leak phenomenon,which is not suitable for detecting the subtle flaws at the irregular surface boundaries.To solve this problem,an efficient intelligent defect detection method based on low contrast image material with non- Gaussian statistics is proposed,which uses the level set to intelligendy segmentthe target object with complex shape and use wavelet transform to extract the defect features.Test results show that the proposed design of the irregular surface defect detection system based on level set and the wavelet transform is effective,which overcomes the residual defects at the edge of detection area.