基于多聚焦图像融合的小孔内表面缺陷检测  

Defect detection of small hole inner surface based on multi-focus image fusion

作　　者：牛群遥 叶明[1] 陆永华[1] 

机构地区：[1]南京航空航天大学机电学院,南京210016

出　　处：《计算机应用》2016年第10期2912-2915,2921,共5页journal of Computer Applications

基　　金：国家自然科学基金资助项目(51575277);南京航空航天大学基本科研业务费专项资金资助项目(NS2014051)~~

摘　　要：针对传统直径3 mm以下小孔内表面缺陷检测缺乏有效检测方法的问题,提出了一种新的基于显微光学与多聚焦图像融合的小孔内表面缺陷检测方法。首先在不同光照环境下,呈斜入射方式沿小孔轴线方向依次采集单侧孔壁的序列图像;然后通过前景光照下得到的掩膜模板提取孔壁图像的感兴趣区域(ROI),并运用快速鲁棒性特征(SURF)算法实现ROI配准;进而采用基于区域清晰度的小波图像融合方法实现ROI多聚焦图像融合;最后以直径2mm喷丝板导孔为实验对象,采用阈值分割算法提取导孔内表面的溶蚀斑进行检测分析。实验结果表明,该方法具有一定的可行性,避免了传统人工检测方法效率低的问题,同时打破了传统直径3 mm以下的小孔内表面检测方法不再适用的局限性。Aiming at the problem of defect defection of the small holes at inner surface with the diameter below 3 mm, a new defection method combining the micro optics and multi-focus image fusion technology was proposed. Firstly, in the different light illumination, the inner surface images of the hole were collected along the axial direction of the hole at oblique incidence. Secondly, the Region Of Interest （ROI） was extracted from the collected hole images based on the mask template obtained by foreground illumination, which was registered through Speeded Up Robust Feature （SURF） algorithm. In the following step, the ROI images were fused through the multi-focus fusion method based on definition of region and wavelet transform. Finally, the guide hole of the spinneret plate with the diameter of 2 mm was used as the experimental object, and the corrosion spots extracted by the segmentation algorithm based on threshold were used for detection and analysis. The research results indicate that the proposed method is certain feasible, which avoids the low efficiency of artificial detection, and breaks the limitation of the traditional defection methods for small holes with diameter below 3 mm.

关 键 词：多聚焦图像融合 小孔 内表面 缺陷检测 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]