This paper presents an efficient method for automatic detection and extraction of blood vessels in retinal images. Specifically, we also delineate vascular intersections/crossovers. The proposed algorithm is composed four steps: matched filtering, local entropy thresholding, length intersection detection. purpose filtering to enhance the vessels. Entropy-based thresholding can well keep spatial structure tree segments. Length used remove misclassified pixels. has been tested on twenty ocular...

This work studies retinal image registration in the context of National Institutes Health (NIH) Early Treatment Diabetic Retinopathy Study (ETDRS) standard. The ETDRS imaging protocol specifies seven fields each retina and presents three major challenges for task. First, small overlaps between adjacent lead to inadequate landmark points feature-based methods. Second, non-uniform contrast/intensity distributions due imperfect data acquisition will deteriorate performance area-based...

Wavelet domain hidden Markov models (HMMs) have been proposed and applied to image processing, e.g., denoising. We develop a new HMM, called local contextual HMM (LCHMM), by introducing the Gaussian mixture field where wavelet coefficients are assumed locally follow distributions determined their neighborhoods. The LCHMM can exploit both statistics intrascale dependencies of at low computational complexity. show that combined with "cycle-spinning" technique achieve state-of-the-art denoising...

In this paper, we propose an articulated and generalized Gaussian kernel correlation (GKC)-based framework for human pose estimation. We first derive a unified GKC representation that generalizes the previous sum of Gaussians (SoG)-based methods similarity measure between template observation both which are represented by various SoG variants. Then, develop (AGKC) integrating kinematic skeleton in multivariate supports subject-specific shape modeling estimation full body hands. further...

In this paper, we propose a three-stream convolutional neural network (3SCNN) for action recognition from skeleton sequences, which aims to thoroughly and fully exploit the data by extracting, learning, fusing inferring multiple motion-related features, including 3D joint positions displacements across adjacent frames as well oriented bone segments. The proposed 3SCNN involves three sequential stages. first stage enriches independently extracted features co-occurrence feature learning....

This paper presents efficient methods for automatic detection and extraction of blood vessels optic disc (OD) both which are two prominent anatomical structures in ocular fundus images. The vessel algorithm is composed four steps, i.e., matched filtering, local entropy-based thresholding, length vascular intersection detection. OD identification involves the active contour (snake) method. Specifically, we use intensity variation contrast to initialize snake model. Simulation results on a set...

We propose a new point set registration method, Global-Local Topology Preservation (GLTP), which can cope with complex non-rigid transformations including highly articulated deformation. The is formulated as Maximum Likelihood (ML) estimation problem two topologically complementary constraints. first the previous Coherent Point Drift (CPD) that encodes global topology constraint by moving one coherently to align second set. second, inspired idea of Local Linear Embedding (LLE), introduced...

Multi-UAV path planning presents a critical challenge in Unmanned Aerial Vehicle (UAV) applications, particularly environments with various obstacles and restrictions. These conditions transform multi-UAV into complex optimization problem multiple constraints, significantly reducing the number of feasible solutions complicating generation optimal flight trajectories. Although slime mold algorithm (SMA) has proven effective missions, it still suffers from limitations such as inadequate...

The feasibility of optical beam-operated magnetic storage systems is assessed. Recent advances in laser technology and transparent materials are described along with methods for writing reading information. In the process, thermomagnetic recording on thin films considered speed resolution calculated, illustrating large performance improvement obtainable compared existing techniques. progress materials, beyond previously used metals, has developed towards more media while retaining intrinsic...

Wavelet-domain hidden Markov models (HMMs), in particular the tree (HMT) model, have been introduced and applied to signal image processing, e.g., denoising. We develop a simple initialization scheme for efficient HMT model training then propose new four-state called HMT-2. find that fits HMT-2 well. Experimental results show performance of denoising using is often improved over two-state developed by Crouse et al. (see ibid., vol.46, p.886-902, 1998).

<para xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> We formulate the issue of joint image segmentation and recognition as an integrated statistical inference problem. A two-layer graphical model is proposed that supports optimal in unified Bayesian framework. Due to explicit modeling two tasks model, efficient non-iterative belief propagation algorithm used for state estimation. The approach applied automatic licence plate (ALPR), it outperforms...

We study automatic target recognition (ATR) in infrared (IR) imagery by applying two recent computer vision techniques, Histogram of Oriented Gradients (HOG) and Bag-of-Words (BoW). propose the idea dense HOG features which are extracted from a set high-overlapped local patches small IR chip we apply vocabulary tree that is learned training images to support efficient scalable BoW-based ATR. develop relevance grouping (RGV) technique improve ATR performance additional voting grouped visual...

State-of-the-art single image dehazing algorithms have some challenges to deal with images captured under complex weather conditions because their assumptions usually do not hold in those situations. In this paper, we develop a deep transmission network for robust dehazing. This simultaneously copes three color channels and local patch information automatically explore exploit haze-relevant features learning framework. We further different structures parameter settings achieve tradeoffs...

The feasibility of a beam-addressable memory using EuO as storage medium and GaAs laser transducer has been discussed in previous paper. In this paper we shall present the results series high-speed, dynamic read/write experiments discuss system implications these experiments. A mechanical scanner was constructed to carry out experiment refrigeration used provide both cooling medium. Writing accomplished by pulsing light spot scanned across film surface. energy density required write 5 μ is...

Single image defogging algorithms based on prior assumptions or constraints have captured much attention because of their simplicity and practicality. However, they still some challenges to deal with foggy images under weather conditions where these may not be effective efficient enough. In this paper, we aim develop a novel algorithm by directly predicting the fog density recovered rather than adopting constraints. order achieve goal, two specific steps are introduced. First, adopt three...

In this paper, a joint multicontext and multiscale (JMCMS) approach to Bayesian image segmentation is proposed. addition the framework, JMCMS applies multiple context models jointly use their distinct advantages, we heuristic multistage, problem-solving technique estimate sequential maximum posteriori of JMCMS. The results on both synthetic mosaics remotely sensed images show that proposed improves classification accuracy, in particular, boundary localization detection over methods using...

Key-frame extraction and object segmentation are usually implemented independently separately due to the fact that they on different semantic levels involve features. In this work, we propose a joint key-frame method by constructing unified feature space for both processes, where is formulated as selection process in context of Gaussian mixture model (GMM)-based video modeling. Specifically, two divergence-based criteria introduced extraction. One recommends leads maximum pairwise interclass...

<para xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> In this paper, we address the problem of target tracking in a collaborative acoustic sensor network. To cope with inherent characteristics and constraints wireless networks, present novel target-tracking algorithm power-aware concerns. The underlying methodology is described as multiple-sensor tracking/fusion technique based on particle filtering. As discussed most recent literature, filtering...

In this paper the first experiment of thermomagnetic writing a hologram on thin film EuO and reconstruction is reported. Writing achieved by using giant pulsed ruby laser as source for thermomagnetization. Erasing accomplished small applied magnetic field. The used because its ability to provide energy required in time that short compared with thermo-relaxation time. A shallow angle (10 μ resolution) was recorded done He–Ne source. formation based fact index refraction material function...

Face images under uncontrolled environments suffer from the changes of multiple factors such as camera view, illumination, expression, etc. Tensor analysis provides a way analyzing influence different on facial variation. However, TensorFace model creates difficulty in representing nonlinearity view subspace. In this paper, to break limitation, we present view-manifold-based (V-TensorFace), which latent manifold preserves local distances multiview face space. Moreover, kernelized...

This paper studies a specific one-class classification problem where the training data are corrupted by significant outliers. Specifically, we interested in support vector machine (OCSVM) approach that normally requires good data. However, perfect usually hard to obtain most real-world applications due inherent variability and uncertainty. To address this issue, propose an OCSVM-based editing method can iteratively purify learn appropriate classifier from trimmed set. The proposed is...