医学图像配准中基于图像梯度的灰度压缩

基于互信息的医学图像配准，其配准精度可以达到亚像素水平，精度高且鲁棒性好，但互信息的巨大计算量使配准速度较慢，不能达到临床使用要求，而互信息的计算速度与图像的灰度阶数有关。为此，针对互信息由于图像灰度级数过多造成互信息计算量大的问题，提出一种基于图像梯度的灰度压缩算法。算法采用图像的梯度信息，根据图像梯度对图像进行非线性灰度映射，同时利用小波对差异图像进行分解和重构。实验结果证明，该算法能减少图像灰度阶数，同时较好地保留图像的细节信息，在保持配准精度的前提下减少配准时间。

Grayscale Compression Based on Image Gradient in Medical Image Registration

The accuracy of medical image registration based on mutual information can reach sub-pixel level,with high accuracy and good robustness.However,frequent involving a huge amount of floating-point calculation,the rate of registration is very slow and can not achieve real-time requirements of clinical use.Because the rate of mutual information is relative to the amount of gray-scales,this paper presents a method based on a nonlinear grayscale mapping method which uses the gradient information of the image.A different image is decomposited and recomposited by a wavelet.Experimental result shows the method can reduce the gray scale of a image at the premise of preserving the image details.Moreover,this algorithm reduces the registration time and maintain the accuracy of registration.