Accurately and reliably obtaining the three-dimensional motion data of individuals in fish schools is not only valuable for fish behaviour analysis and hydrodynamics studies but also be helpful in areas such as bio-inspired robot design. Video tracking is the most effective gateway to obtain the quantitative motion data of continuously moving objects. In this paper we propose a method for obtaining the quantitative three-dimensional trajectory of individuals in fish schools. The proposed method works on videos captured by multiple synchronized cameras with the help of a suggested three-camera imaging system. The proposed method follows the master-slave paradigm in which it tracks fish in the master view with the help of the convolutional neural network in the first, and then associates each certain fish in the master view with detections in the slave view by formulating the cross-view data association as the moment-wise linear assignment problems. Experiments have conducted on public datasets to completely evaluate the performance. The proposed method outperforms other state-of-the-art methods.

As a routine tool for screening and examination, CT plays an important role in disease detection and diagnosis. Real-time table removal in CT images becomes a fundamental task to improve readability, interpretation and treatment planning. Meanwhile, it makes data management simple and benefits information sharing and communication in picture archiving and communication system. In this paper, we proposed an automated framework which utilized parallel programming to address this problem. Eight full-body CT images were collected and analyzed. Experimental results have shown that with parallel programming, the proposed framework can accelerate the patient table removal task up to three times faster when it was running on a personal computer with four-core central processing unit. Moreover, the segmentation accuracy reaches 99% of Dice coefficient. The idea behind this approach refreshes many algorithms for real-time medical image processing without extra hardware spending.

Image sharpness is one of the most determining factors for image readability and scene understanding. How to accurately quantify it is a hot topic. This paper systematically validates a previously proposed index for full-reference image sharpness assessment (edge preservation ratio, EPR). Based on Gaussian blurring images in LIVE, CSIQ, TID2008 and TID2013 databases, we firstly evaluated EPR accuracy on five edge detectors on LIVE and selected an optimal one for further analysis. Then nine state-of-the-art image quality assessment metrics are compared, including full-reference, no-reference and dedicated image sharpness assessment categories. Experimental results demonstrate (1) Canny is an optimal edge detector for EPR implementation; (2) EPR is a top-ranking image sharpness assessment metric that outperforms PSNR and SSIM and rivals FSIM; and (3) EPR accords more closely with human subjective judgment than involved image sharpness assessment metrics. This study also indicates that image sharpness assessment is still full of challenges and utilizing deep learning architectures to learning the direct mapping from images to quality will be a trend in the near future.

Fetal magnetic resonance imaging has been widely used for brain malformation assessment, brain growth estimation and related disease diagnosis. However, how to estimate fetal brain motion is pending which hampers image interpretation. This paper presents an image registration method to address this problem. It takes total-variation regularization (TV) to model fetal brain motions and adopts a penalty toward piece-wise convergence. Based on four clinical cases, the proposed method was verified from perceived visual perception and landmark errors, and also compared to nonlinear Levenberg-Marquart least square optimization (L2). Visual perception indicates that TV-based registration outperforms L2-based method with less tissue difference, and landmark errors show that the landmark displacement decreases from 4.86±1.76mm to 0.82±0.51mm with L2-based method and to 0.42±0.57mm with TV-based method. One-way ANOVA verifies that both L2- and TV-based methods significantly reduce landmark errors (p<;0.0001), while TV-based method outperforms L2-based method (p<;0.0065). This paper provides a feasible solution to estimate fetal brain motions and benefits brain development study and disease diagnosis.