IVR image enhancement with ICA domain shrinkage filter and multi-scale filter

本研究项目主要关注IVR图像质量改善和船舶增强。针对泊松降噪和血管提取，我们开发了一种自适应ICA收缩滤波器和多尺度滤波器，然后将这两种滤波器融合，得到具有鲜明血管特征和低噪声效果的高质量IVR图像。因此，我们希望将开发的算法应用于实际的IVR图像增强，用于诊断应用。主要研究成果如下：(1)开发了用于IVR图像船舶提取的多尺度滤波器，并探索了高速算法；(2)提出了一种自适应ICA收缩滤波器对IVR图像进行泊松降噪，然后融合两种算法对IVR图像进行增强；(3)将我们开发的算法应用于实际的IVR图像进行诊断应用；(4)采用DVBV方法对不同算法增强后的IVR图像进行定量评价。评价结果表明，该算法增强后的图像具有最高的DV/BV率。

This research project mainly focus on IVR image quality improvement and vessel enhancement. We developed an adaptive ICA shrinkage filter and multi-scale filter for Poisson noise reduction and vessel extraction, and then fusion these two developed filters to achieve high quality IVR images with distinctive vessel and less noise effectiveness. Therefore, we want to apply the developed algorithms for real IVR image enhancement for diagnosis application. The research achievements are as the following :(1) Developed multi-scale filter for Vessel extraction from IVR images, and explore the high-speed algorithm;(2) Proposed an adaptive ICA shrinkage filter for Poisson noise reduction in IVR images, and then fusion the two developed algorithm for IVR image enhancement;(3) Applied our developed algorithm to real IVR images for diagnosis application;(4) Adopted DVBV methods for quantitative evaluation of the enhanced IVR images by different algorithms. Evaluation results showed that the enhanced image by our proposed algorithm has the highest DV/BV rate.