Affiliations: [a] Shandong University of Finance and Economics, Jinan, Shandong, China | [b] Digital Media Technology Key Lab of Shandong Province, Jinan, Shandong, China | [c] Qianfoshan Hospital of Shandong Province,Jinan, Shandong, China | [d] Shandong University, Jinan, Shandong, China
Correspondence:
[*]
Corresponding author: Hui Liu, Shandong University of Finance and Economics, Jinan, Shandong, China. Tel.: +86 1528 8857 221; E-mail:liuh_lh@sdufe.edu.cn
Abstract: Medical CT imaging has an important sense in the
treatment process. However, the low resolution of CT images could easily
affect the final diagnosis, which is influenced by the resolution and the
radiation dosage. We propose to solve this problem by using an adaptive
image super-resolution reconstruction algorithm. First, the algorithm of the
CT image of quad-tree decomposition obtains adaptive access to different
scales of the image patches. Then, we exploit K-means clustering algorithm
to determine the cluster center. Using the center of cluster, we can obtain
the mapping function between the low-resolution image patches and the
high-resolution image patches. Finally, the algorithm reconstructs a
high-resolution image through the mapping function. The experimental results
have shown that the proposed method is capable of enhanced CT image
reconstruction, peak signal-to-noise ratio (PSNR) and structural similarity (SSIM).
Keywords: Image reconstruction, image super resolution, quad-tree decomposition, cluster center, self-adaption
