摘要
为了解决医学MR图像边缘模糊、亮度不合理和噪声抑制效果不好等问题,提出了一种灰度插值直方图线性化算法。该方法先对含混合噪声的医学MR图像进行直方图线性化处理,抑制部分噪声,使亮度达到合理的范围,进一步将得到的结果进行多项式插值运算,消除象素块间的灰度差异,使图像达到层次清晰。仿真实验结果表明,该算法获得的图像亮度、峰值信噪比的值和结构相似性指数指标均最大,其中,峰值信噪比值高出其他方法约为1.6~3.2 dB,结构相似性指数指标高出其他方法约为5%~7%。该方法可有效地降低噪声,较好地保持了医学MR图像边缘和细节信息,其效果明显优于自适应直方图均衡化和直方图局部线性化等方法。
In order to solve the problems of blurred edges,unreasonable brightness and poor noise suppression of medical MR image,a gray interpolation histogram linearization algorithm is proposed.This method firstly linearizes the histogram of the medical MR image with mixed noise,so as to removing part of the noise and make the brightness reach a reasonable range,and then performs polynomial interpolation operation on the obtained results to eliminate the gray difference between pixel blocks and make the image clear.The simulation results show that the brightness,peak signail-to-noise ratio(PSNR)and struetural similarity index method(SSIM)of the image obtained by this algorithm are the largest,and the PSNR is about 1.6~3.2 dB higher than other denoising methods,and the SSIM is about 5%~7%higher.This algorithm can effectively reduce the noise and keep the edge and detail information of medical MR image,and its effect is obviously better than that of adaptive histogram equalization and histogram local linearization.
作者
陈军
CHEN Jun(Department of Medical Education,Dingxi Campus,Gansu University of Chinese Medicine,Dingxi 743000,China)
出处
《贵州大学学报（自然科学版）》
2024年第3期49-53,77,共6页
Journal of Guizhou University:Natural Sciences
基金
甘肃省科技计划资助项目(20JR10RA327)。
关键词
灰度插值
直方图
医学MR图像
线性化
gray interpolation
histogram
medical MR image
linearization
