基于小波变换模极大值的医学图像融合技术

针对目前的融合算法不能有效解决强去噪能力和细节信息保留之间的矛盾,提出一种基于小波变换模极大值特征的窗口区域强度自适应加权平均融合算法.首先,用选定的小波基提取不同尺度下小波变换模极大值特性;然后,利用信号与噪声的Lipschitz指数在局部奇异处呈现不同的表现形式的特性,滤除噪声信号;接着,计算模极大值的局部区域强度,动态地实现在不同尺度的子图像的小波分解系数之间分配权重;最后,将处理后的子图像重建得到融合后的图像.通过对CT和PET图像的融合实验,证明了该方法既可以适应不同特征图像的融合任务,又能达到有效抑制噪声而保留尽可能多的细节信息的目的.

Fusion Technology of Medical Image Based on Wavelet Transform Modulus Maximum

Aim at the incapable of solving the contradiction between the strong de-noise ability and the detail information reservation of actual fusion algorithms, a new algorithm based on the characteristics of wavelet transform modulus maximum is presented in this paper. The algorithm is studied using the wavelet coefficients adaptive weighted averaging based on windows-region intensity. Firstly, The characteristics of wavelet transform modulus maximum in different scales is extracted with chosen wavelet radix. Secondly, noise is filtered by using the different characteristics of Lipschitz index between the signal and noise in local singularity points. Thirdly, local region intensity of modulus maximum is computed, then the weight of wavelet coefficients is distributed dynamically in sub-images of different scales. Lastly, the fusion image is completed by reconstructing sub-images. The fusion experiments using CT and PET are also done. The results show that this method can adapt itself to various fusion demand with different images characteristics, and can be capable to restrain noise and reserve detail information to a great extent.