基于差分曲率的各项异性扩散小波图像降噪算法研究

在$\gamma $放射性内污染修正均匀冗余阵列编码成像中，为了进一步减少使用最大似然估计(Maximum Likelihood Expectation Maximisation，MLEM)算法进行编码图像重建的伪影，并且提高重建的效率，本文提出了基于差分曲率的各项异性扩散小波图像降噪(Anisotropic Diffusion Wavelet Image Denoising based on Differential Curvature, 简称ADWIDDC)算法, 结合MLEM算法得到了改进的应用于编码成像的MLEM-ADWIDDC算法。该算法首先采用MLEM算法对探测器得到的投影数据进行图像重建，再利用互补成像方法消减近场伪影，最后采用ADWIDDC算法进一步降低图像伪影。模拟结果表明，对于131I源呈圆环分布的编码图像，以重建算法运行时间为195 s为截止条件，MLEM-ADWIDDC算法能够更好地去除伪影，重建的$\gamma $放射源图像纹理也更加清晰；对于131I源呈“CDUT”分布的编码图像，以重建图像信噪比达到5.10 dB为截止条件，本文的MLEM-ADWIDDC算法运行时间为98.36 s，比仅采用MLEM算法消耗的时间缩短了49%。

Research on Anisotropic Diffusion Wavelet Image Denoising Algorithm Based on Differential Curvature

In modified uniformly redundant arrays code imaging of gamma radiation internal contamination, in order to further decrease reconstruction artifacts with Maximum Likelihood Expectation Maximisation(MLEM) algorithm and to improve the efficiency of reconstruction, this paper proposed an anisotropic diffusion wavelet image denoising algorithm based on differential curvature (ADWIDDC algorithm), and combining with MLEM algorithm, a modified code imaging MLEM-ADWIDDC algorithm was obtained. Firstly, the reconstruction of the projection data obtained from the detector was conducted with MLEM algorithm. Secondly, the near-field artifacts were decreased with complementary imaging method. Finally, the image artifacts were further reduced by ADWIDDC algorithm. The simulation results show that the MLEM-ADWIDDC algorithm can better remove the artifacts and the texture of the reconstructed gamma source image is clearer when 131I source is designed as circular distribution, with the cut-off time of 195 s for the reconstruction algorithm to run. When 131I source is designed as “CDUT” distribution and the signal-to-noise ratio is set 5.10 dB as cut-off condition of code imaging reconstruction, the real reconstruction time of the MLEM-ADWIDDC algorithm is 98.36 s, which is 49% shorter than that of MLEM algorithm alone.