【摘要】 挠性电路板(FPC)的图像采集与预处理是自动检测的基础。针对检测缺陷不同提出3种照明方案,对欲检测缺陷拥有较好成像效果,若检测所有缺陷,可采用2种方案结合分别采集图片;只有在同轴光照明下才可准确识别FPC折痕缺陷,为区别折痕与表面曲向,研究FPC表面曲向消除方法(CSE),根据FPC基材表面曲向区域挠度引起的灰度值渐变情况,以最小像素个数、最大灰度梯度、均方差为筛选条件,识别图像中的表面曲向区域,以表面曲向区域中的灰度极小值为真实灰度值,消除图像中的表面曲向。利用高斯滤波与CSE结合对图像进行预处理,图像信噪比比高斯滤波提升5.80 d B。更多还原

【Abstract】 Method of image acquisition and preprocessing for flexible printed circuit board( FPC) is the foundation of automatic detection. Aiming at the difference of defects three kinds of illuminant is presented,which have good imaging effect. If all defects need to detect,combine two kinds of illuminant could be adopted. Only in coaxial illumination,FPC defects could be identified accurately. The method of FPC curved surface eliminating( CSE) is researched in order to distinguish defects. According to the gray value gradient in image,which caused by regional deflection,for the minimum number of pixels,the maximum gray gradient,mean variance as screening conditions,curved surface region in the image could be identified. The gray of curved surface region is replaced by the minimum gray of the region,the curved surface could be eliminated. Combine the Gaussian filter with CSE as the preprocessing,image SNR promotes 5. 80 d B than Gauss filter.更多还原