Search

In computer vision, accurately estimating a 3D human skeleton from a single RGB image remains a challenging task. Inspired by the advantages of multi-view approaches, we propose a method of predicting enhanced 2D skeletons (specifically, predicting the joints' relative depths) from multiple virtual viewpoints based on a single real-view image. By fusing these virtual-viewpoint skeletons, we can then estimate the final 3D human skeleton more accurately. Our network consists of two stages. The first stage is composed of a two-stream network: the Real-Net stream predicts 2D image coordinates and the relative depth for each joint from the real viewpoint, while the Virtual-Net stream estimates the relative depths in virtual viewpoints for the same joints. Our network's second stage consists of a depth-denoising module, a cropped-to-original coordinate transform (COCT) module, and a fusion module. The goal of the fusion module is to fuse skeleton information from the real and virtual viewpoints so that it can undergo feature embedding, 2D-to-3D lifting, and regression to an accurate 3D skeleton. The experimental results demonstrate that our single-view method can achieve a performance of 45.7 mm on average per-joint position error, which is superior to that achieved in several other prior studies of the same kind and is comparable to that of other sequence-based methods that accept tens of consecutive frames as the input.

Objectives: Colistin-induced nephrotoxicity prolongs hospitalisation and increases mortality. The study aimed to construct machine learning models to predict colistin-induced nephrotoxicity in patients with multidrug-resistant Gram-negative infection., Methods: Patients receiving colistin from three hospitals in the Clinical Research Database were included. Data were divided into a derivation cohort (2011-2017) and a temporal validation cohort (2018-2020). Fifteen machine learning models were established by categorical boosting, light gradient boosting machine and random forest. Classifier performances were compared by the sensitivity, F1 score, Matthews correlation coefficient (MCC), area under the receiver operating characteristic (AUROC) curve, and area under the precision-recall curve (AUPRC). SHapley Additive exPlanations plots were drawn to understand feature importance and interactions., Results: The study included 1392 patients, with 360 (36.4%) and 165 (40.9%) experiencing nephrotoxicity in the derivation and temporal validation cohorts, respectively. The categorical boosting with oversampling achieved the highest performance with a sensitivity of 0.860, an F1 score of 0.740, an MCC of 0.533, an AUROC curve of 0.823, and an AUPRC of 0.737. The feature importance demonstrated that the days of colistin use, cumulative dose, daily dose, latest C-reactive protein, and baseline haemoglobin were the most important risk factors, especially for vulnerable patients. A cutoff colistin dose of 4.0 mg/kg body weight/d was identified for patients at higher risk of nephrotoxicity., Conclusions: Machine learning techniques can be an early identification tool to predict colistin-induced nephrotoxicity. The observed interactions suggest a modification in dose adjustment guidelines. Future geographic and prospective validation studies are warranted to strengthen the real-world applicability., (Copyright © 2024 The Author(s). Published by Elsevier Ltd.. All rights reserved.)

This paper presents an RGB-NIR (Near Infrared) dual-modality technique to analyze the remote photoplethysmogram (rPPG) signal and hence estimate the heart rate (in beats per minute), from a facial image sequence. Our main innovative contribution is the introduction of several denoising techniques such as Modified Amplitude Selective Filtering (MASF), Wavelet Decomposition (WD), and Robust Principal Component Analysis (RPCA), which take advantage of RGB and NIR band characteristics to uncover the rPPG signals effectively through this Independent Component Analysis (ICA)-based algorithm. Two datasets, of which one is the public PURE dataset and the other is the CCUHR dataset built with a popular Intel RealSense D435 RGB-D camera, are adopted in our experiments. Facial video sequences in the two datasets are diverse in nature with normal brightness, under-illumination (i.e., dark), and facial motion. Experimental results show that the proposed method has reached competitive accuracies among the state-of-the-art methods even at a shorter video length. For example, our method achieves MAE = 4.45 bpm (beats per minute) and RMSE = 6.18 bpm for RGB-NIR videos of 10 and 20 s in the CCUHR dataset and MAE = 3.24 bpm and RMSE = 4.1 bpm for RGB videos of 60-s in the PURE dataset. Our system has the advantages of accessible and affordable hardware, simple and fast computations, and wide realistic applications.

In this paper, a multi-focus image stack captured by varying positions of the imaging plane is processed to synthesize an all-in-focus (AIF) image and estimate its corresponding depth map. Compared with traditional methods (e.g., pixel- and block-based techniques), our focus-based measures are calculated based on irregularly shaped regions that have been refined or split in an iterative manner, to adapt to different image contents. An initial all-focus image is first computed, which is then segmented to get a region map. Spatial-focal property for each region is then analyzed to determine whether a region should be iteratively split into sub-regions. After iterative splitting, the final region map is used to perform regionally best focusing, based on the Winner-take-all (WTA) strategy, i.e., choosing the best focused pixels from image stack. The depth image can be easily converted from the resulting label image, where the label for each pixel represents the image index from which the pixel with the best focus is chosen. Regions whose focus profiles are not confident in getting a winner of the best focus will resort to spatial propagation from neighboring confident regions. Our experiments show that the adaptive region-splitting algorithm outperforms other state-of-the-art methods or commercial software in synthesis quality (in terms of a well-known Q metric), depth maps (in terms of subjective quality), and processing speed (with a gain of 17.81~40.43%).

A robust lensless optical data embedding system using both the computer-generated concealogram and the digital Fresnel hologram (DFH) is proposed. The concealogram and the DFH are cascaded as the input and filter planes, respectively. When the concealogram is illuminated by a plane wave, the hidden data can be extracted at the output plane without using any lenses. The longitudinal positions of the filter and the output planes, as well as the wavelength, can be used as the secret keys to enhance system security. Furthermore, the robustness of the proposed system against noise and distortion is also demonstrated., (© 2011 Optical Society of America)

What we believe to be a new technique, based on a modified Gerchberg-Saxton algorithm (MGSA) and a phase modulation scheme in the Fresnel-transform domain, is proposed to reduce cross talks existing in multiple-image encryption and multiplexing. First, each plain image is encoded and multiplexed into a phase function by using the MGSA and a different wavelength/position parameter. Then all the created phase functions are phase modulated to result in different shift amounts of the reconstruction images before being combined together into a single phase-only function. Simulation results show that the cross talks between multiplexed images have been significantly reduced, compared with prior methods [Opt. Lett.30, 1306 (2005); J. Opt. A8, 391 (2006)], thus presenting high promise in increasing the multiplexing capacity and encrypting grayscale and color images.

A novel fast double-phase retrieval algorithm for lensless optical security systems based on the Fresnel domain is presented in this paper. Two phase-only masks are efficiently determined by using a modified Gerchberg-Saxton algorithm, in which two cascaded Fresnel transforms are replaced by one Fourier transform with compensations to reduce the consumed computations. Simulation results show that the proposed algorithm substantially speeds up the iterative process, while keeping the reconstructed image highly correlated with the original one.

For object segmentation, traditional snake algorithms often require human interaction; region growing methods are considerably dependent on the selected homogeneity criterion and initial seeds; watershed algorithms, however, have the drawback of over segmentation. A new downstream algorithm based on a proposed extended gradient vector flow (E-GVF) field model is presented in this paper for multiobject segmentation. The proposed flow field, on one hand, diffuses and propagates gradients near object boundaries to provide an effective guiding force and, on the other hand, presents a higher resolution of direction than traditional GVF field. The downstream process starts with a set of seeds scored and selected by considering local gradient direction information around each pixel. This step is automatic and requires no human interaction, making our algorithm more suitable for practical applications. Experiments show that our algorithm is noise resistant and has the advantage of segmenting objects that are separated from the background, while ignoring the internal structures of them. We have tested the proposed algorithm with several realistic images (e.g., medical and complex background images) and gained good results.