基于梯度的低对比度X线图像分割方法

Method for Low Contrast X-ray Image Segmentation Based on Gradient

在医学骨折诊断领域,X线图像是骨折诊断的重要手段,X线图像中骨骼组织的精确分割为后续的诊断提供了重要依据。为此,文中首先利用Sobel边缘检测算子得到X线图像的梯度图,然后利用灰度图像单目标跟踪算法,提出了一种基于梯度的多目标分割算法,经过三轮跟踪获得了骨骼组织的精确定位,实现了医学骨科X线图像的骨骼分割。最后,将分割结果与边缘检测和阈值方法的分割结果进行了比较。试验结果表明,该方法对低对比度图像能够得到好的分割效果,比较适合医学骨科X线图像的骨骼分割。

X-ray images are important means for fracture diagnosis in the field of medical fracture diagnosis. The accurate segmentation of skeletal tissues from X-ray image provides a basis for future diagnosis In this paper, the gradient image of X-ray image based on Sobel edge detection operator is obtained first, and then, according to the tracing algorithm of single objective, a multi-objective segmentation algorithm based on gradient is proposed. The multi-objective segmentation algorithm is used in gray image to precisely locate skeletal tissues after three-time tracings and implement skeleton segmentation of medicine orthopedics X-ray image. Finally, this paper compares the result with that of edge detection and threshold method. The experiment shows that this method can get good segmentation results for low-contrast images and is suitable for skeleton segmentation of medicine orthopedics X-ray image .