APP下载

基于潜在低秩表示的双模态医学图像融合

2021-02-28刘海涛龙黄枭刘珊姜瀚周涛

电脑知识与技术 2021年35期

刘海涛 龙黄枭 刘珊 姜瀚 周涛

摘要:医学图像融合是图像处理科研领域的热门课题。针对传统的PET-CT图像融合方法有效地融合了CT图像和PET图像的互补信息,但是在图像整体细节信息保留能力仍然不足,现提出一种基于潜在低秩表示的双模态医学图像融合方法,首先,将医学原图像通过潜在低秩表示方法分解为低秩部分、显著部分以及噪声部分。在低秩部分中,采用加权平均融合规则进行融合,以此来保留更多的图像整体信息,在显著部分中,使用求和策略进行融合,保留图像的细节信息。然后,将这两部分的融合图像再进行重构,得到最终融合图像。实验结果表明,融合图像无论是从客观上还是主观上,都具有良好的医学观察效果。

关键词:医学图像;图像融合;潜在的低秩表示;加权平均策略;低秩分解

中图分类号:TP318      文献标识码:A

文章编号:1009-3044(2021)35-0005-04

Bimodal Medical Image Fusion Based on Latent Low Rank Representation

LIU Hai-tao1, LONG Huan-xiao1, LIU Shan1, JIANG Han1, ZHOU Tao1,2

(1.School of Computer Science and Engineering, North Minzu University, Yinchuan 750021, China;2. Key Laboratory of the State Ethnic Affairs Commission for Intelligent Processing of Image and Graphics,North Minzu University, Yinchuan 750021, China)

Abstract: Bimodal medical image fusion is a hot topic in the field of image processing. The traditional pet CT image fusion method effectively integrates the complementary information of CT image and PET image, but the overall detail information retention ability of the image is still insufficient. Now, a new Bimodal medical image fusion method based on the latent low rank representation is proposed. Firstly, the medical source images are decomposed into low rank part, significant part and noise part by latent low rank representation. In the low rank part, the weighted average strategy is adopted to preserve more overall information of the image. In the the significant part,the summation strategy is used to preserve detailed information of the image. Then the fusion images of these two parts are reconstructed to obtain the final fusion image. The experimental results show that the fusion image has good medical observation effect both objectively and subjectively.

Key words: medical image; image fusion; latent low-rank representation;weighted average strategy; low rank decomposition

圖像融合是从两幅或多幅图像的同一区域提取互补信息,通过融合方法融合到一幅图像中,有效地克服单一传感器的局限性和差异性[1]。在医学影像技术中,由于CT成像是通过扫描横断面,得到断面图像,可以有效地消除器官和组织平面图像投影重叠的缺点。CT图像根据正常器官和病变器官的密度不同来进行识别病灶,其器官组织图像分辨率极高但对病灶本身的显示效果不佳。PET图像是利用放射性核素标记原理,通过该物质在生物体中的移动聚集情况,反映生命体功能活动。但是PET图像在成像过程中由于受到康普顿效应、散射等衰减因素的影响,采集得到的图像数据和病状体的实际情况不符,图像质量失真,分辨率低。……

登录APP查看全文