摘要
在X射线成像中,由于探测器尺寸限制,对于大尺寸部位拍摄时需要多次分段拍摄,为了自动将这些影像无缝地拼接在一起,提出了多尺度的自动融合拼接算法。首先对输入影像进行低通滤波预处理以尽量减少噪声带来的不利影响,然后通过智能人体部位判断,对于不同部位采用提取不同特征区域,并对其进行了针对性的增强处理,然后进行多尺度由粗到精的平移估计,最后通过灰度分析,在全局进行补偿灰度差,在重叠区进行灰度融合,实现无缝拼接。实验及临床应用表明此拼接算法快速可靠,拼接图像质量良好。
Long bones such as spine, upper and lower extremities should be taken into several fragment images clue to the limited size of deteetor in X-ray imaging system. For the sake of stitching these individual images seamlessly, a multiscale automatic stitching algorithm is presented in this paper. The paper describes the whole process： first, low-pass filtering is taken to the original X-ray images to reduce the noises, and then the feature area is extracted according to the intellectual body part judgment, multiscale matching procedure is performed to estimate the shift in both horizontal and vertical direction, finally, intensity eompensation to the whole image and intensity fusion to the overlap area are taken to realize seamless stitching. Lots of experimental results and clinic applications show that the method is fast and reliable with excellent image quality.
出处
《中国医疗设备》
2008年第7期12-15,共4页
China Medical Devices
关键词
X线透视
图像拼接
多尺度匹配
灰度融合
DR
X-ray fluoroscopy
image stitching
muhiscale matching
intensity fusion
DR
