摘要
在骨龄自动评价系统研究过程中,对X射线图像中手骨兴趣区域进行定位是一份艰巨的工作。为了能准确定位手骨兴趣区域(ROI),提出一种先用改进的脉冲耦合神经网络(PCNN)对手腕骨进行二值分割,然后根据手骨特征信息提取指骨边缘并对其进行曲线拟合,接着在原图中沿曲线用Sobel算子定位到手骨ROI的方法。定位准确率为93.9%的实验结果表明该方法是有效的。
In the research process of automatic bone age evaluation system, locating the ROI of hand bones in X ray image is a difficult task. In order to accurately locate the hand bones ROI, we propose such a method, which first uses the improved pulse coupled neural network （PCNN） to carry out binary segmentation on hand-twist bone, and then extracts the phalange edges according to hand bones feature information and makes curve fitting on them, the next it locates to the hand bone ROI along the curves in primary image with Sobel operator. The experimental results with locating accuracy rate being 93.9% show that the method is effective.
出处
《计算机应用与软件》
CSCD
北大核心
2014年第7期226-228,232,共4页
Computer Applications and Software
关键词
脉冲耦合神经网络
骨龄
兴趣区域
图像分割
手骨
Pulse coupled neural network
Bone age
Region of interest （ROI）
Image segmentation
Hand bone
