摘要
针对当前医学图像配准方法存在配准精度和效率低的问题,提出基于混合策略的数字化医学影像技术下多模态图像配准方法。将多模态图像分量划分为低频子带分量与高频子带分量,对图像中低频分量实行多尺度的Retinex增强,完成图像低频子带去噪。利用阈值法对图像各个尺度与方向上高频噪声进行抑制,实现高频子带去噪。对去噪后的低频子带和高频子带实行模糊对比度增强操作,得到图像最终增强结果。将图像增强结果引入图像配准,利用力矩主轴法对图像进行粗匹配。以粗匹配为基础,通过Powell算法对最佳配准变换参数进行求解,采用最佳配准变换参数实现精图像配准。仿真结果表明,所提方法配准效率高,配准精度最高为98%。上述方法在图像配准精确性和配准耗时上与当前方法相比均有所提升,可为医疗领域的发展提供一定程度上的支撑。
Current medical image registration method has low accuracy and efficiency of registration.This article proposes a method for multi-modality image registration in digital medical imaging technology based on mixed strategy.At first,the .components of multi -modality image were divided into low-frequency sub-band components and high-frequency sub-band components.Secondly,multi-scale Retinex enhancement was performed on the low- frequency components in image to complete the noise reduction of low-frequency sub-band in image.Moreover, threshold method was used to suppress the high-frequency noise in each scale and direction of image,so as to achieve the noise reduction of high-frequency sub-band.Then,fuzzy contrast enhancement was performed on the low frequency sub-band and high-frequency sub-band after the noise reduction to obtain the final enhancement result of image.In addition,the image enhancement result was introduced into image registration,and the principal-axis method was used for coarsely matching image.Based on the rough matching,the Powell algorithm was used to obtain the best registration transformation parameter.Finally,the best registration transformation parameter was used to achieve fine registration of image.Simulation results show that the proposed method has high registration efficien- cy.The maximum accuracy of registration is 98%.Compared with the current method,the proposed method improves the accuracy of image registration and reduces the time consumption of registration,which can provide a certain support for the development of medical field.
作者
何锡嘉
凌巍高
张雅欣
梁志胜
HE Xi-jia;LING Wei-gao;ZHANG Ya-xin;LIANG Zhi -sheng(Information and Management School,Guangxi Medical University,Nanning Guangxi 530021,China)
出处
《计算机仿真》
北大核心
2018年第12期166-170,共5页
Computer Simulation
基金
广西高校中青年教师基础能力提升项目(2018KY0097)
关键词
数字化医学影像
多模态图像
配准
Digital medical imaging
Multi-modality image
Registration