李肖肖, 聂仁灿, 周冬明, 谢汝生. 图像增强的拉普拉斯多尺度医学图像融合算法[J]. 云南大学学报(自然科学版), 2019, 41(5): 908-917. doi: 10.7540/j.ynu.20180840
引用本文: 李肖肖, 聂仁灿, 周冬明, 谢汝生. 图像增强的拉普拉斯多尺度医学图像融合算法[J]. 云南大学学报(自然科学版), 2019, 41(5): 908-917. doi: 10.7540/j.ynu.20180840
LI Xiao-xiao, NIE Ren-can, ZHOU Dong-ming, XIE Ru-sheng. Laplacian multi-scale medical image fusion algorithm for image enhancement[J]. Journal of Yunnan University: Natural Sciences Edition, 2019, 41(5): 908-917. DOI: 10.7540/j.ynu.20180840
Citation: LI Xiao-xiao, NIE Ren-can, ZHOU Dong-ming, XIE Ru-sheng. Laplacian multi-scale medical image fusion algorithm for image enhancement[J]. Journal of Yunnan University: Natural Sciences Edition, 2019, 41(5): 908-917. DOI: 10.7540/j.ynu.20180840

图像增强的拉普拉斯多尺度医学图像融合算法

Laplacian multi-scale medical image fusion algorithm for image enhancement

  • 摘要: 针对不同临床诊断的需求产生了多种类型的医学图像,如:核磁共振成像(MRI),正电子发射断层扫描图像(PET),单光子发射计算机断层扫描技术(SPECT). 单一功能的医学成像技术往往不能更好地诊断出病人的病情,利用多模态医学图像融合技术获得的图像具有较强的解剖学意义,与原始单扫描图像相比具有较高的光谱信息. 首先,将基于像素级的图像增强技术与拉普拉斯金字塔分解方案相结合,使分解的残差子带图像(RSI)具有更清晰的图像纹理特征,与此同时保持基底子带图像(BSI)更平滑. 然后,对RSI图像采用基于兴趣信息的图像融合规则获取高频系数融合图像;对BSI图像采用局部能量最大值方案的融合规则获取较平滑的金字塔顶层融合图像. 最后,利用拉普拉斯金字塔逆运算重建融合图像. 实验结果表明融合图像保留了图像边缘信息的同时增强了图像细节信息,具有更好的视觉效果,客观评价指标QMIQIFCQVIFQMEQSDQUQIQSSIM也取得了较大的提升.

     

    Abstract: Various types of medical images are generated for different clinical diagnostic needs. Such as magnetic resonance imaging (MRI), positron emission tomography (PET), and single-photon emission computed tomography (SPECT). Single function medical imaging technology is difficult to accurately diagnose the disease. Therefore, multi-modal medical image fusion technology can obtain images with robust anatomical significance and high spectral information. First, combining the image enhancement technique based on pixel processing with the Laplace pyramid decomposition scheme, which can obtain the residual sub-band image (RSI) with clearer image texture features and the base sub-band image (BSI) with smoother. Then, the high-frequency coefficient fused image is obtained using the image fusion rule based on the interest information for the RSI image. And a smoother pyramid top-level fused image is obtained by using the fusion rule of the local energy maximum scheme for the BSI image. Finally, the inverse Laplace pyramid transformation is utilized to reconstruct the fusion image. The experimental results show that the fused image preserves image edge information, enhances image detail information for better visual effects and objective evaluation metrics have more advantages including QMI, QIFC, QVIF, QME, QSD, QUQI, QSSIM.

     

/

返回文章
返回