Search

This study evaluates if hand movements, tracked using digital video, can quantify in-context surgical performance. Participants of varied experience completed simple interrupted suturing and running subcuticular suturing tasks. Marker-less motion tracking software traced the two-dimensional position of a region of the hand for every video frame. Four expert observers rated 219 short video clips of participants performing the task from 0 to 10 along the following visual analog scales: fluidity of motion, motion economy, tissue handling, and coordination. Expert ratings of attending surgeon hand motions (mean=7.5, sd=1.3) were significantly greater ( p

Objective: A method for automatically classifying lifting postures from simple features in video recordings was developed and tested. We explored if an “elastic” rectangular bounding box, drawn tightly around the subject, can be used for classifying standing, stooping, and squatting at the lift origin and destination. Background: Current marker-less video tracking methods depend on a priori skeletal human models, which are prone to error from poor illumination, obstructions, and difficulty placing cameras in the field. Robust computer vision algorithms based on spatiotemporal features were previously applied for evaluating repetitive motion tasks, exertion frequency, and duty cycle. Methods: Mannequin poses were systematically generated using the Michigan 3DSSPP software for a wide range of hand locations and lifting postures. The stature-normalized height and width of a bounding box were measured in the sagittal plane and when rotated horizontally by 30°. After randomly ordering the data, a classification and regression tree algorithm was trained to classify the lifting postures. Results: The resulting tree had four levels and four splits, misclassifying 0.36% training-set cases. The algorithm was tested using 30 video clips of industrial lifting tasks, misclassifying 3.33% test-set cases. The sensitivity and specificity, respectively, were 100.0% and 100.0% for squatting, 90.0% and 100.0% for stooping, and 100.0% and 95.0% for standing. Conclusions: The tree classification algorithm is capable of classifying lifting postures based only on dimensions of bounding boxes. Applications: It is anticipated that this practical algorithm can be implemented on handheld devices such as a smartphone, making it readily accessible to practitioners.

An active multi-debris removal mission planning algorithm for low Earth orbit (LEO) debris removal is presented. We assume that a space platform carrying multiple debris-deorbiting nano-satellites will be launched to carry out the debris-removal mission. This platform travels from one debris object to the next using a drift-orbit-transfer strategy. Each debris is assigned a reward value reflecting potential collision risk reductions if the debris is deorbited. The objective of mission planning is to determine an ordering of debris to be deorbited that maximizes the total reward due to debris removal subject to the constraints of (a) the total amount of velocity change for orbit transfer, (b) the total duration of the debris removal mission, and (c) the number of deorbiting nano-satellites carried on board the platform. In this work, a new approach is taken to approximate the response curve of drift-orbit transfers between each debris pair using a numerically interpolated continuous curve. An optimal orbit transfer strategy then is adopted to minimizes the sum of relative Δ v consumption and relative transfer duration. The task of multi-debris removal mission planning is formulated as a combinatorial tree search discrete optimization problem using this optimal orbit transfer cost estimate. A greedy heuristic with polynomial time complexity is proposed to select the next debris to deorbit that maximizes the reward-to-cost ratio. The effectiveness of this novel approach is demonstrated using real-world Iridium 33 debris cloud data. It is observed that this algorithm delivers superior performance while consuming a small fraction of computing time compared to state of the art mission planning algorithms.

In this paper, we introduce a novel multi-view image denoising algorithm using 3D focus image stacks (3DFIS) to exploit image redundancy within and across views. Robust disparity map is first estimated using the 3DFIS with texture-based view selection and patch-size variation scheme. Leveraging both 3DFIS and the estimated disparity map, the proposed algorithm effectively denoises the target view from multiple views through a low rank minimization approach that incorporates robust similarity metrics and occlusion handling techniques. The paper combs through a number of existing image denoising methods, including the preliminary results in our earlier research efforts, and then details the ways, means and merits of our proposed algorithm. With extensive experiments, we conclude that this novel algorithm is superior over various existing state-of-the-art approaches in terms of both visual and quantitative performance.

An image compression and encryption method integrating compressive sensing (CS) and Detour Cylindrical-diffraction-based double Random-phase Encoding (DCRE) is proposed. The plaintext image is first compressed by CS and then encrypted by the DCRE algorithm. The existing integrated CS and double random phase encoding (DRPE) methods are known to be vulnerable to the plaintext attack and even ciphertext-only attack. The combination of CS and DCRE, however, will be free of the plaintext attack and ciphertext-only attack, free of phase retrieval attack and allows enlarged key space. Numerical simulation results demonstrate the effectiveness and flexibility of the proposed scheme.

Moving target tracking in directional sensor networks has recently attracted attention by right of special directional sensing features. Unlike omnidirectional sensors, the directional sensor senses the target only in the direction of its orientation. It can provide quantized direction that indicates the presence or absence of the target in the sensing field, rather than just the analog measurement of sensing signal with respect to the detected target. A quantized Kalman filter (QKF) based on both quantized directions and analog ranging measurements is derived in the minimum mean-square error (MMSE) sense. Its performances of mean square estimation error (MSE) and complexity are also analyzed. Then, a reduced-complexity QKF of high-accuracy is pursued to facilitate its implementation. It is proved that the QKF yields a smaller MSE than the traditional extended Kalman filter (EKF) merely based on analog measurements. The posterior Cram $\acute{e}$ r-Rao lower bound (PCRLB) is introduced as the performance measure. The performance advantages of the proposed QKF are demonstrated using Monte Carlo simulations in a target tracking application using ultrasonic ranging sensors.

A phase-retrieval attack free cryptosystem based on the cylindrical asymmetric diffraction and double-random phase encoding (DRPE) is proposed. The plaintext is abstract as a cylinder, while the observed diffraction and holographic surfaces are concentric cylinders. Therefore, the plaintext can be encrypted through a two-step asymmetric diffraction process with double pseudo random phase masks located on the object surface and the first diffraction surface. After inverse diffraction from a holographic surface to an object surface, the plaintext can be reconstructed using a decryption process. Since the diffraction propagated from the inner cylinder to the outer cylinder is different from that of the reversed direction, the proposed cryptosystem is asymmetric and hence is free of phase-retrieval attack. Numerical simulation results demonstrate the flexibility and effectiveness of the proposed cryptosystem.

Recently, color image encryption has attracted extensive research interests. However, challenging issues, such as linearity, complex-valued output, and information disclosure, remain. To address these issues, we propose an asymmetric color image cryptosystem by using detour cylindrical-diffraction and phase reservation and truncation (PRT). In this cryptosystem, a complex combination of two color components of a color image is encrypted through a detour cylindrical-diffraction process with double random phase masks, which are located on the input and intermediate surfaces. The diffraction distribution of complex amplitude is separated into phase and amplitude parts using PRT. The phase parts are saved as asymmetric keys, while the two amplitude parts are combined with another color component and saved as ciphertext, respectively. The asymmetric detour cylindrical-diffraction destroys the linearity of the proposed cryptosystem and solves the issue of information disclosure. Meanwhile, the PRT ensures that the cryptosystem is asymmetric and has amplitude-only output, which further destroys the linearity. Therefore, the three issues in color image encryption are resolved. Numerical simulation results validate the effectiveness and flexibility of the proposed cryptosystem.

A novel joint sparse representation based multi-source localization method is presented in this work. With the proposed approach, the direction of arrival (DOA) of each source may be formulated as a sparse reconstruction problem using spatial array covariance matrix, and the corresponding source locations can be formulated as a joint sparse representation of array covariance matrices (JSRACM) for multiple phase arrays. The proposed approach transforms the source location estimation problem into a spatial sparse signal representation (SSSR) optimization problem, which avoids the conventional DOA association in the case of multiple sources. To mitigate the high computation complexity of the JSRACM approach, a novel binary sparse indicative vector (SIV) is introduced to represent the support of joint SSSR of array covariance matrices. As such, the multiple source locations may be estimated by solving an unconstrained optimization problem of the SIV vector using existing FOCUSS-like algorithms. The resulting SIV-JSRACM algorithm does not require prior information of the number of sources nor initial source location estimates. Simulation results demonstrate the advantages of the proposed approach.

Cylindrical computer-generated hologram is a promising approach to realize a display with 360°field of view. However, conventional cylindrical hologram employs an inside-out propagation model and suffers from two main drawbacks: limited object size and lack of effective reconstructed method. Previously, we proposed to fix these problems using an outside-in propagation model with reversed propagation direction of the inside-out model. We also derived corresponding diffraction calculation formula for the outside-in propagation model. In this work, we investigate a non-constant obliquity factor in the outside-in propagation model, and show that it is the projection of the unit complex amplitude in the propagation direction onto the outer normal of the observation point. We then propose to apply fast Fourier transform to accelerate the convolution operation needed for diffraction calculation. We conducted experiments on inverse diffraction and reconstruction of the cylindrical objects. Very encouraging results demonstrate the validity of this proposed approach.

We propose a novel random triggering-based modulated wideband compressive sampling (RT-MWCS) method to facilitate efficient realization of sub-Nyquist rate compressive sampling systems for sparse wideband signals. Under the assumption that the signal is repetitively (not necessarily periodically) triggered, RT-MWCS uses random modulation to obtain measurements of the signal at randomly chosen positions. It uses multiple measurement vector method to estimate the nonzero supports of the signal in the frequency domain. Then, the signal spectrum is solved using least square estimation. The distinct ability of estimating sparse multiband signal is facilitated with the use of level triggering and time-to-digital converter devices previously used in random equivalent sampling scheme. Compared to the existing compressive sampling (CS) techniques, such as modulated wideband converter (MWC), RT-MWCS is with simple system architecture and can be implemented with one channel at the cost of more sampling time. Experimental results indicate that, for sparse multiband signal with unknown spectral support, RT-MWCS requires a sampling rate much lower than Nyquist rate, while giving great quality of signal reconstruction.

We investigate the problem of sparse signal medium access control in the wireless sensor networks. The proposed sparse signal Aloha (SSA) transmits sampled compressive data to the fusion center using Aloha random access protocol. In order to maximize the overall data transmission rate in the presence of packet collision, redundant data are randomly subsampled at individual sensor nodes according to an optimal subsampling rate before transmission. Data loss due to random subsampling and collision in the wireless channel are compensated using compressive sensing recovery method at the fusion center. Simulation results verify the theoretical analysis and show the significant improvement of SSA compared with existing CS-based random access method.

We previously developed a method for classifying lifting postures using dimensions of a rectangular bounding box drawn tightly around the subject for a single camera view that is more tolerable of conditions encountered in industrial settings than high precision tracking and can be practically implemented on a smart hand-held device. This study explores the use of simple bounding box dimensions to predict trunk angle while lifting. Mannequin poses were generated using the Michigan 3DSSPP software for 105 postures across six anthropometries. A regression model for predicting trunk angle was created (adjusted R2=0.91, p2 = 0.80). This algorithm should be useful for calculating trunk kinematic properties that are associated with increased risks of low-back disorders including trunk speed and acceleration using successive video frames of predicted trunk flexion angles.

In this paper, we propose a novel multi-view image denoising algorithm based on convolutional neural network (MVCNN). Multi-view images are arranged into 3D focus image stacks (3DFIS) according to different disparities. The MVCNN is trained to process each 3DFIS and generate a denoised image stack that contains the recovered image information for regions of particular disparities. The denoised image stacks are then fused together to produce a denoised target view image using the estimated disparity map. Different from conventional multi-view denoising approaches that group similar patches first and then perform denoising on those patches, our CNN-based algorithm saves the effort of exhaustive patch searching and greatly reduces the computational time. In the proposed MVCNN, residual learning and batch normalization strategies are also used to enhance the denoising performance and accelerate the training process. Compared with the state-of-the-art single image and multi-view denoising algorithms, experiments show that the proposed CNN-based algorithm is a highly effective and efficient method in Gaussian denoising of multi-view images.

We formulate the active multi-debris removal mission planning task as a Reinforcement-Learning (RL) problem and developed an adjusted Deep Q-Learning (DQN) solution. We propose novel definitions of the state space, action sets, and rewards in the context of active multi-debris removal mission planning. These definitions facilitate recasting the mission planning problem into a RL problem. As such, a powerful DQN algorithm may be applied to solve the mission planning problem using an RL approach. We test this new approach using a subset of Iridium 33 debris cloud. Very encouraging results are observed. Future applications to a reactive autonomous space mission planner are also discussed.

Objective: This study explores how common machine learning techniques can predict surgical maneuvers from a continuous video record of surgical benchtop simulations. Background: Automatic computer vision recognition of surgical maneuvers (suturing, tying, and transition) could expedite video review and objective assessment of surgeries. Method: We recorded hand movements of 37 clinicians performing simple and running subcuticular suturing benchtop simulations, and applied three machine learning techniques (decision trees, random forests, and hidden Markov models) to classify surgical maneuvers every 2 s (60 frames) of video. Results: Random forest predictions of surgical video correctly classified 74% of all video segments into suturing, tying, and transition states for a randomly selected test set. Hidden Markov model adjustments improved the random forest predictions to 79% for simple interrupted suturing on a subset of randomly selected participants. Conclusion: Random forest predictions aided by hidden Markov modeling provided the best prediction of surgical maneuvers. Training of models across all users improved prediction accuracy by 10% compared with a random selection of participants. Application: Marker-less video hand tracking can predict surgical maneuvers from a continuous video record with similar accuracy as robot-assisted surgical platforms, and may enable more efficient video review of surgical procedures for training and coaching.

We propose a high-performance visual target tracking (VTT) algorithm based on classified-patch kernel particle filter (CKPF). Novel features of this VTT algorithm include sparse representations of the target template using the label-consistent K-singular value decomposition (LC-KSVD) algorithm; Gaussian kernel density particle filter to facilitate candidate template generation and likelihood matching score evaluation; and an occlusion detection method using sparse coefficient histogram (ASCH). Experimental results validate superior performance of the proposed tracking algorithm over state-of-the-art visual target tracking algorithms in scenarios that include occlusion, background clutter, illumination change, target rotation, and scale changes.

In this paper, an asymmetric multiple-image cryptosystem is proposed using LUP decomposition (LUPD) in the spherical diffraction (SpD) domain. To the best of our knowledge, the SpD transform is applied to the optical encryption for the first time. In the proposed cryptosystem, multiple grayscale images are encoded into a sparse and binary amplitude distribution with the cascaded LUPD and SpD transform. The matching private keys are generated by LUPD, which is convenient for administrative permissions by setting different numbers of private keys. Since LUPD and SpD are both non-linear processes, high security can be achieved to resist attacks such as the retrieval attack. In addition, up to 12 plaintext images can be encrypted with high decryption quality using the proposed cryptosystem. On the other side, the sparse and binary ciphertext can be stored or transmitted with 1/8 space compared with general 8-bit-grayscale ciphertext, which is a merit of using LUPD. The feasibility and effectiveness of the proposed asymmetric cryptosystem are verified by numerical simulation results.

A novel network-on-chip (NoC) integrated congestion control and flow control scheme, called Network-Cognitive Traffic Control (NCogn.TC), is proposed. This scheme is cognizant of the fluidity levels in on-chip router buffers and it uses this measurement to prioritize the forwarding of flits in the buffers. This preferential forwarding policy is based on the observation that flits with higher levels of fluidity are likely to arrive at their destinations faster, because they may require fewer routing steps. By giving higher priority to forward flits in high-fluidity buffers, scarce buffer resources may be freed-up sooner in order to relieve on-going traffic congestion. In this work, a buffer cognition monitor is developed to rapidly estimate the buffer fluidity level. An integrated congestion control and flow control algorithm is proposed based on the estimated buffer fluidity level. Tested with both synthetic traffic patterns as well as industry benchmark traffic patterns, significant performance enhancement has been observed when the proposed Network-Cognitive Traffic Control is compared against conventional traffic control algorithms that only monitor the buffer fill level.

A one-bit joint sparse representation direction of arrival (OBJSR-DOA) estimation approach is proposed in this letter. By exploiting the joint spatial and spectral correlations inherent in acoustic sensor array data, the proposed OBJSR-DOA approach provides reliable DOA estimation from only the sign bit of randomly subsampled acoustic sensor data. The random subsampling and single-bit quantization allow significant reduction of data to be transmitted to the fusion center without additional energy consumption requirement in the source coding/compression operation. Compared with existing compressive sensing-based DOA estimation methods, the superiority of the proposed approach in providing data volume reduction and performance improvement is verified by both simulations and field experiments using a prototype wireless sensor array network platform.

The authors present a novel cardinalised probability hypothesis density (CPHD) algorithm under the assumption of a glint measurement noise model with unknown inverse covariance. Noise parameters are assumed to have a Gamma prior distribution so that the predicted and updated PHDs can have mixture of Gaussians representations. A variational Bayesian expectation maximisation procedure is applied to iteratively estimate parameters of the mixture distributions through random hypersurface model CPHD prediction and update steps. Simulation results show that the proposed algorithm can adaptively track extended objects with unknown object number and glint measurement noise, while achieving higher precision compared against the traditional approach.

A novel real-time task allocation and scheduling scheme is proposed for a multi-core system incorporated in a Bidirectional Network-on-Chip (BiNoC) platform. Given a task graph, this scheme seeks to minimize the total execution time by allocating ready-to-execute tasks to as many available cores as possible subject to the real-time deadlines of each task. A refinement process is introduced to update the priority ranking of a task list so as to meet the timing constraints. In particular, the communication overhead is considered by incorporating the packet routing paths and delays into the overall optimization process. In doing so, the flexibility of bidirectional links of BiNoC is exploited to alleviate traffic congestion, such that more tasks could be executed concurrently at different cores and overall execution time be reduced. To validate the effectiveness of this proposed scheme, extensive simulations have been performed. The results clearly demonstrate the superior performance of this proposed scheme compared to existing approaches that did not exploit the flexibility of BiNoC.

A novel L - infinity event-triggered networked control scheme under time-varying communication delay is proposed. The control objective is to maintain L - infinity input to state stability while minimizing triggering and communication overhead under uncertainties. The event triggering is asynchronous for satisfying each subsystem performance requirement individually, and the event triggering strategy is adaptive to the variable communication delay to reduce the communication cost without degrade of system performance (input to state stability). The relations between the communication delay and the event triggering conditions are investigated and related bounds are derived analytically. An example of electrical power system frequency control task is used to illustrate the effectiveness of the proposed scheme in terms of achieving identical control performance with lower communication overhead.

Two motion models are proposed to enhance the performance of video object tracking (VOT) algorithms. The first one is a random walk model that captures the randomness of motion patterns. The second one is a data-adaptive vector auto-regressive (VAR) model that exploits more regular motion patterns. The performance of these models is evaluated empirically using real-world datasets. Three real-time publicly available visual object trackers: the normalized cross-correlation (NCC) tracker, the New Scale Adaptive with Multiple Features (NSAMF) tracker, and the correlation filter neural network (CFNet) are modified using each of these two models. The tracking performances are then compared against the original formulation. It is observed that both models of the prior information lead to performance enhancement of all three trackers. This validates the hypothesis that when training videos are available, prior information embodied in the motion models can improve the tracking performance.

Existing laparoscopic surgery systems use a single laparoscope to visualize the surgical area with a limited field of view (FoV), necessitating maneuvering the laparoscope to search a target region. In some cases, the laparoscope needs to be moved from one surgical port to another one to detect target organs. These maneuvers would cause longer surgical time and degrade the efficiency of operation. We hypothesize that if an array of cameras can be deployed to provide a stitched video with an expanded FoV and small blind spots, the time required to perform multiple tasks at different sites can be significantly reduced. We developed a micro-camera array that can enlarge the FoV and reduce blind spots between the cameras by optimizing the angle of cameras. The video stream of this micro-camera array was designed to be processed in real-time to provide a stitched video with the expanded FoV. We mounted this micro-camera array to a Fundamentals of Laparoscopic Surgery (FLS) laparoscopic trainer box and designed an experiment to validate the hypothesis above. Surgeons, residents, and a medical student were recruited to perform a modified bean drop task, and the completion time was compared against that measured using a traditional single-camera laparoscope. It was observed that utilizing the micro-camera array, the completion time of the modified bean drop task was 203 &plusmn, 55 s while using the laparoscope, the completion time was 245 &plusmn, 114 s, with a p-value of 0.00097. It is also observed that the benefit of using an FoV-expanded camera array does not diminish for subjects who are more experienced. This test provides convincing evidence and validates the hypothesis that expanded FoV with small blind spots can reduce the operation time for laparoscopic surgical tasks.

A method is introduced for automatically classifying lifting postures using simple features obtained through drawing a rectangular bounding box tightly around the body on the sagittal plane in video recordings. Mannequin postures were generated using the University of Michigan 3DSSPP software encompassing a variety of hand locations and were classified into squatting, stooping, and standing. For each mannequin posture a rectangular bounding box was drawn tightly around the mannequin for views in the sagittal plane and rotated by 30 o horizontally. The bounding box dimensions were measured and normalized based on the standing height of the corresponding mannequin. A classification and regression tree algorithm was trained using the height and width of the bounding box to classify the postures. The resulting algorithm misclassified 0.36% of the training-set cases. The algorithm was tested on 30 lifting postures collected from video recordings a variety of industrial lifting tasks, misclassifying 3.33% of test-set cases. The sensitivity and specificity, respectively were 100.0% and 100.0% for squatting, 90.0% and 100.0% for stooping, and 100.0% and 95.0% for standing. The algorithm was capable of classifying lifting postures based only on dimensions of bounding boxes which are simple features that can be measured automatically and continuously. We have developed computer vision software that continuously tracks the subject’s body and automatically applies the described bounding box.

A novel frame-subsampled, drift-resilient (FSDR) video object tracking (VOT) algorithm is proposed. Two design goals are sought: to improve the accuracy and to reduce the processing time. The drifting problem is mitigated with a drift detector and accompanying drift recovery mechanism. When a drift is detected, the recovery mechanism provides an opportunity to put the tracking back on track. To gather context-dependent statistics required for these procedures, an initial short segment of the video sequence will be used as a training sequence. Thus, this algorithm is more suitable for video sequences much longer than several minutes. To reduce computing time, a novel frame-subsampling strategy is proposed to process the VOT on small subset of frames. The trajectory of the tracked object on frames that are skipped will be estimated via interpolation. Compared with state of art VOT algorithms, dramatic improvement of performance (accuracy) and orders of magnitude computing time reduction are observed.

This paper reviews the 2nd NTIRE challenge on single image super-resolution (restoration of rich details in a low resolution image) with focus on proposed solutions and results. The challenge had 4 tracks. Track 1 employed the standard bicubic downscaling setup, while Tracks 2, 3 and 4 had realistic unknown downgrading operators simulating camera image acquisition pipeline. The operators were learnable through provided pairs of low and high resolution train images. The tracks had 145, 114, 101, and 113 registered participants, resp., and 31 teams competed in the final testing phase. They gauge the state-of-the-art in single image super-resolution.

A simplified frequency-interleaved framework with digital back-end correction is presented for wideband analog-to-digital converters. Instead of channel-by-channel signal recovery, an iterative structure is proposed to simultaneously compensate the mismatches including in-phase/quadrature (I/Q) imbalances, jitter-induced distortions, as well as overlapping effects due to imperfect channel separation and incomplete anti-aliasing. The effectiveness of this novel architecture is validated via extensive simulations.

This paper investigates the cross-layer design problem with the goal of maximizing the network lifetime for energy-constrained underwater wireless sensor networks (UWSNs). We first jointly consider link scheduling, transmission power and transmission rate in a proposed optimization problem with the adoption of time division multiple access (TDMA) schedules. Then, we propose an iterative algorithm to solve the optimization problem. It alternates between (1) link scheduling and (2) computation of transmission powers and transmission rates. In fact, the convergence of such iterative algorithm can be mathematically and empirically supported. We evaluate our algorithm for several network topologies. Extensive simulation results demonstrate the superiority of the proposed approach.

Performance-cost trade-offs in video object tracking tasks for long video sequences is investigated. A novel frame-subsampled, drift-resilient (FSDR) video object tracking algorithm is presented that would achieve desired tracking accuracy while dramatically reducing computing time by processing only sub-sampled video frames. A new pattern matching score metric is proposed to estimate the probability of drifting. A drift-recovery procedure is developed to enable the algorithm to recover from a drift situation and resume accurate tracking. Compared against state-of-the-art video object tracking algorithms, dramatic performance (accuracy) enhancement and cost (computing time) reduction are observed.

We propose an adaptive visual target tracking algorithm based on Label-Consistent K -Singular Value Decomposition (LC-KSVD) dictionary learning. To construct target templates, local patch features are sampled from foreground and background of the target. LC-KSVD then is applied to these local patches to simultaneously estimate a set of low-dimension dictionary and classification parameters (CP). To track the target over time, a kernel particle filter (KPF) is proposed that integrates both local and global motion information of the target. An adaptive template updating scheme is also developed to improve the robustness of the tracker. Experimental results demonstrate superior performance of the proposed algorithm over state-of-art visual target tracking algorithms in scenarios that include occlusion, background clutter, illumination change, target rotation and scale changes.

A robust multi-view disparity estimation algorithm for noisy images is presented. The proposed algorithm constructs 3D focus image stacks (3DFIS) by projecting and stacking multi-view images and estimates a disparity map based on the 3DFIS. To make the algorithm robust to noise and occlusion, a texture-based view selection and patch size variation scheme based on texture map is proposed. Experiment results indicate that the proposed algorithm outperforms conventional stereo matching algorithms as well as previously reported multi-view disparity estimation algorithms under noisy conditions.

A unified and accurate fast diffraction calculation between a pair of concentric cylindrical surfaces is proposed. Analysis of the obliquity factor shows that the physical meaning of it is the projection of the unit complex amplitude in the propagation direction onto the outer normal of the observation point. Therefore, a unified and accurate diffraction calculation formula for both inside-out and outside-in propagation models is achieved. Furthermore, it is accurately accelerated by a fast calculation method using fast Fourier transformation with a speed-up over 3.9×106, which is compared with the direct calculation method under resolution of 512×512 pixels. Through analysis and comparison, it is found that the computing error is significant in the approximate method of the inside-out propagation model, which releases the limitation of the ratios of the outer radius and the height to the inner radius of the cylinders. The simulation results are presented to verify the correctness and accuracy of our proposal.

In this paper, we consider the task of stitching video frames captured by an array of micro-cameras in order to create an enlarged field of view video for surgical applications. We propose a method for mitigating the problem of parallax when dealing with non-planar scenes with moving objects that cross boundaries of adjacent views, causing discontinuities at view boundaries of the stitched video. The discontinuities created by parallax are especially noticeable when the object has long straight lines that cross the boundary. In this work, we propose a method of aligning the edges of objects affected by the parallax in the scene using epipolar geometry and then applying secondary transformations to correct parallax discontinuities. Our method leverages the regular shape of the object and is suitable for surgical applications where few if any feature points may be extracted from the appearance of the objects.

Reliable data transmission over lossy communication link is expensive due to overheads for error protection. For signals that have inherent sparse structures, compressive sensing (CS) is applied to facilitate efficient sparse signal transmissions over lossy communication links without data compression or error protection. The natural packet loss in the lossy link is modeled as a random sampling process of the transmitted data, and the original signal will be reconstructed from the lossy transmission results using the CS-based reconstruction method at the receiving end. The impacts of packet lengths on transmission efficiency under different channel conditions have been discussed, and interleaving is incorporated to mitigate the impact of burst data loss. Extensive simulations and experiments have been conducted and compared to the traditional automatic repeat request (ARQ) interpolation technique, and very favorable results have been observed in terms of both accuracy of the reconstructed signals and the transmission energy consumption. Furthermore, the packet length effect provides useful insights for using compressed sensing for efficient sparse signal transmission via lossy links.