当前位置: X-MOL 学术Pattern Recognit. Image Anal. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Human Visual System Consistent Model for Wireless Capsule Endoscopy Image Enhancement and Applications
Pattern Recognition and Image Analysis ( IF 0.7 ) Pub Date : 2020-09-15 , DOI: 10.1134/s1054661820030219
V. B. Surya Prasath , Dang N. H. Thanh , Le Thi Thanh , N. Q. San , S. Dvoenko

Abstract

Visualization of inner gastrointestinal (GI) tract is an important aspect in diagnosis of diseases such as the bleeding and colon cancer. Wireless capsule endoscopy (WCE) provides painless imaging of the GI tract without much discomfort to patients via near-lights imaging model and with burst light emitting diodes (LEDs). This imaging system is designed to minimize battery power and the capsule moves through the GI tract with natural peristalsis movement and the color video data are captured via wireless transmitter in the WCE. Despite the advantages of WCE videos, the obtained frames exhibit uneven illumination and sometimes result in darker regions that may require enhancement afterwards for better visualization of regions of interest. In this work, we extend a human visual system (HVS) based image enhancement model that uses a feature-linking neural network model based on timing precisely of the spiking neurons. Experimental results on various WCE frames show that we can obtain better enhancement of regions of interest and compared to other enhancement approaches in the literature we obtain better quality restorations in general. Further, we show the applicability of our enhancement method on improving the automatic image segmentation, and 3D shape from shading visualization reconstruction indicating that it is viable to be used within a computer-aided diagnosis systems for GI tract diseases.


中文翻译:

无线胶囊内窥镜图像增强的人类视觉系统一致性模型及其应用

摘要

内胃肠道的可视化是诊断诸如出血和结肠癌等疾病的重要方面。无线胶囊内窥镜检查(WCE)可通过近光成像模型和爆裂发光二极管(LED)为胃肠道提供无痛成像,而不会给患者带来很大不适。该成像系统旨在最大程度地减少电池电量,并且胶囊以自然的蠕动运动在胃肠道中移动,并且彩色视频数据通过WCE中的无线发射器捕获。尽管有WCE视频的优点,但获得的帧显示出不均匀的照明,有时会导致较暗的区域,之后可能需要进行增强以更好地显示感兴趣的区域。在这项工作中 我们扩展了基于人类视觉系统(HVS)的图像增强模型,该模型使用了基于尖峰神经元精确定时的特征链接神经网络模型。在各种WCE框架上的实验结果表明,我们可以获得感兴趣区域的更好增强,并且与文献中的其他增强方法相比,我们通常可以获得更好的质量恢复。此外,我们展示了我们的增强方法在改善自动图像分割方面的适用性,以及从阴影可视化重建中获得的3D形状,表明该方法可用于胃肠道疾病的计算机辅助诊断系统。在各种WCE框架上的实验结果表明,我们可以获得感兴趣区域的更好增强,并且与文献中的其他增强方法相比,我们通常可以获得更好的质量恢复。此外,我们展示了我们的增强方法在改善自动图像分割方面的适用性,以及从阴影可视化重建中获得的3D形状,表明该方法可用于胃肠道疾病的计算机辅助诊断系统。在各种WCE框架上的实验结果表明,我们可以获得感兴趣区域的更好增强,并且与文献中的其他增强方法相比,我们通常可以获得更好的质量恢复。此外,我们展示了我们的增强方法在改善自动图像分割方面的适用性,以及从阴影可视化重建中获得的3D形状,表明该方法可用于胃肠道疾病的计算机辅助诊断系统。
更新日期:2020-09-15
down
wechat
bug