摘要
基于锥束CT的印刷电路板(PCB)无损检测是近年来出现的新的PCB检测手段,但是CT图像中普遍存在的金属伪影容易导致PCB图像的错误分割。为了抑制金属伪影对PCB图像分割的影响,文章提出了一种基于图割的交互式分割方法。利用受金属伪影影响的PCB图像灰度分布特点设计了新的权值函数,并进一步利用了用户输入的信息,将用户选取的种子点信息作为计算区域项权重的重要参数。通过对多组印刷电路板图像进行分割实验比较,结果表明与经典的Graph Cuts和Grabcut等算法相比,文章方法在准确性方面有明显提高。
Non-destructive testing using cone beam computed tomography （CBCT） is a new method for printed circuit board （PCB） inspection, but a large number of artifacts in CT image may lead to false segmentation which is still a challenging problem. To deal with it, a new interactive segmenta- tion method based on graph cuts is proposed in this paper. According to the histogram of the PCB image,a new cost function is designed and seed points information is used as a parameter for region cost at the same time which results in a better use of user input information. Comparison between different PCB image segmentation experiments demonstrate that the proposed method have better per- formance compared with the state-of-art methods such as Graph Cuts and Grabcut in terms of seg- mentation accuracy.
出处
《信息工程大学学报》
2013年第1期67-71,共5页
Journal of Information Engineering University
基金
国家863计划资助项目(2012AA011603)
关键词
印刷电路板
无损检测
图割
伪影
CT图像分割
PCB
non-destructive testing
graph cuts
artifacts
CT image segmentation
