基于非下采样Shearlet变换耦合能量关联度的医学图像融合算法

Medical Image Fusion Algorithm Based on Nonsubsampled Shearlet Transform Coupled with Energy Correlation Degree

摘要:

为了克服当前较多医学图像融合方法在采用图像的能量信息融合图像时,忽略了不同图像能量的关联度,使得融合结果存在细节丢失现象和模糊现象等问题,提出了一种非下采样 变换 耦合能量关联度的医学图像融合算法。借助 变换,在多尺度下对输入医学图像进行解析,获取其低频及高频子带系数。以图像的能量信息为依据,构造能量关联度函数,测量不同图像的关联程度。根据不同图像的关联度,设计不同的低频子带融合规则,获取信息含量丰富且连贯性较好的融合低频子带。在空间频率函数的基础上,注入图像的对角信息,使之成为多元空间频率函数,以计算图像的清晰度。引入标准差函数,计算图像的对比度。联合图像的清晰度和对比度信息,获取纹理及对比度等特征都较优良的融合高频子带。基于逆 变换,重构融合结果。主观和客观实验结果表明,较当前较为流行的医学图像融合技术而言,所提方法具备更高的融合质量,呈现出更多的纹理细节和更高的清晰度。

Abstract:

In order to overcome the problem that many medical image fusion methods ignore the correlation degree of different image energy when they use the energy information of image to fuse images, which makes the fusion results have the phenomenon of loss of detail and ambiguity, a medical image fusion algorithm based on non subsampled shearlet transform coupled with energy correlation degree is proposed. With the help of NSST transform, the input medical image is analyzed in multi-scale to obtain its low-frequency and high-frequency subband coefficients. Based on the energy information of the image, the energy correlation function is constructed to measure the correlation degree of different images. According to the correlation degree of different images, different fusion rules of low-frequency subbands are designed to obtain fusion low-frequency subbands with rich information and good coherence. On the basis of spatial frequency function, the diagonal information of the image is injected to make it a multivariate spatial frequency function to calculate the clarity of the image. The standard deviation function is introduced to calculate the contrast of the image. Combined with the clarity and contrast information of the image, the fusion high-frequency subband with better texture and contrast characteristics is obtained. Based on the inverse NSST transform, the fusion result is reconstructed. Subjective and objective experimental results show that compared with the current popular medical image fusion technology, the proposed method has higher fusion quality, more texture details and higher definition.