All-sky 1 km land surface temperature (LST) data are urgently needed. Two widely applied approaches to derive such LST merging thermal infrared remote sensing (TIR)–passive microwave (PMW) observations and TIR reanalysis data. However, as only the Moderate Resolution Imaging Spectroradiometer (MODIS) is adopted source for merging, current all-sky products limited MODIS observation time. Therefore, a gap still remains in terms of with higher temporal resolution or at other times (e.g.,...

A heterogeneous network of workstations (NOW) introduces a new performance factor into distributed computing: large variation the computing power different workstations. This unique makes traditional models/metrics for homogeneous measurement and evaluation not suitable computing. We present models which quantify heterogeneity networks characterize effects. The consider effects both time-sharing in nondedicated environment. Speedup, efficiency scalability are defined. These general enough to...

Modern Internet streaming services have utilized various techniques to improve the quality of media delivery. Despite characterization access patterns and user behaviors in many measurement studies, few studies focused on themselves, particularly experiences they offer end users resources systems that consume. In order gain insights into current thus provide guidance designing resource-efficient high systems, we collected a large workload from thousands broadband home business hosted by...

In general, the performance of parallel graph processing is determined by three pairs critical parameters, namely synchronous or asynchronous execution mode (Sync Async), Push Pull communication mechanism (Push Pull), and Data-driven Topology-driven traversing scheme (DD TD), which increases complexity sophistication programming system implementation GPU. Existing graph-processing frameworks mainly use a single combination in entire for given application, but we have observed their variable...

Image segmentation is critical and challenging in computer vision medical image analysis. Despite decades of research, existing algorithms are still subject to typical problems, such as over-segmentation, under-segmentation, non-closed spurious edges. In this paper, taking the carpal bones from hand X-ray images foreground regions, we propose a approach integrate segmentations region-based boundary-based methods tackle these problems. First, adaptive local thresholding Canny edge detection...

We released open-source software Hadoop-GIS in 2011, and presented published the work VLDB 2013. This initiated development of a new spatial data analytical ecosystem characterized by its large-scale capacity both computing storage, high scalability, compatibility with low-cost commodity processors clusters software. After more than decade research development, this has matured is now serving many applications across various fields. In paper, we provide background on why started project give...

During quality-assurance procedures in the mass production of small-sized curved optical lenses, fine defects are usually detected via manual observation, which is not recommended owing to associated drawbacks high error rate, low efficiency, and nonamenability quantitative analysis. To address this concern, paper presents a comprehensive defect-detection system based on transmitted fringe deflectometry, dark-field illumination, light transmission. Experimental results obtained study reveal...

3D topography metrology of optical micro-structured surfaces is critical for controlled manufacturing and evaluation properties. Coherence scanning interferometry technology has significant advantages measuring surfaces. However, the current research faces difficulties designing high accuracy efficient phase shifting, characterization algorithms surface metrology. In this paper, parallel unambiguous generalized phase-shifting T-spline fitting are proposed. To avoid ambiguity improve...

An edge detection process in computer vision and image processing detects any types of significant features appearing as discontinuities intensities. This paper presents our experience with parallelizing an application algorithm that reduces noise unnecessary detail a gray-scale from coarse level to fine resolution by using focusing technique. Numerical methods parallel implementations are presented. The algorithms implemented on three representative message-passing architectures: low-cost...

Liver segmentation from abdominal Computer Tomography (CT) images plays an important role in liver disease diagnosis as well surgical planning. In this paper, a hybrid approach is proposed for fully automatic position search and CT images. First intensity range detected based on prior knowledge of volume. Then region interest (ROI) extracted using atlas-based affine non-rigid registration. At the last step, to achieve more accurate segmentation, major tumors are gray level distance...

In order to improve the denoising quality of pulsar signal, an empirical mode decomposing method (EMD) signal based on cell proportion shrinking is proposed. Firstly, decomposed into a series intrinsic functions (IMF), and part between two adjacent zero-crossing within IMF defined as cell. Then, optimal proportional factor constructed by treating basic unit analysis. Finally, all cells are denoised shrinking, model established. The experimental results show that compared with EMD algorithms...

Medical image based biomarkers are being established for therapeutic cancer clinical trials, where assessment is among the essential tasks. Large scale often performed by a large group of experts retrieving images from centralized repository to workstations markup and annotate images. In such environment, it critical provide high performance management system that supports efficient concurrent retrievals in distributed environment. There several major challenges: throughput data over...

Segmentation and other image processing operations rely on convolution calculations with heavy computational memory access demands. The article presents an analysis of a texture segmentation application containing 96/spl times/96 convolution. Sequential execution required several hours single processor systems over 99% the time spent performing large 70% to 75% is attributable cache misses within We implemented same CM-5, iPSC/860 PVM distributed multicomputers, tailoring parallel algorithms...

Network latency, the delay caused by communication between processors and memory modules over network in a multiprocessor system, is major source of degraded parallel computing performance. We first give an overview experimental metric which uses latency to measure evaluate scalability programs architectures. put emphasis on evaluation sources their methods program execution. report results evaluating several scientific algorithms KSR-1. In comparison, we also present preliminary experiments...

Stereo matching and semantic segmentation are significant tasks in binocular satellite 3D reconstruction. However, previous studies primarily view these as independent parallel tasks, lacking an integrated multitask learning framework. This work introduces a solution, the Single-branch Semantic Network (S3Net), which innovatively combines stereo using Self-Fuse Mutual-Fuse modules. Unlike preceding methods that utilize or disparity information independently, our method dentifies leverages...

We describe a programmable and scalable Convolutional Neural Network (CNN) hardware accelerator optimized for mobile edge inference computing. The is comprised of 4 heterogeneous engines - input engine, filter post processing output engine. specialized execute independently concurrently. All have core set common instructions with each engine further specific functions. the operation provide silicon validated results number CNN networks including LeNet-5, TinySSD, SqueezeNet. blind modulation...

Sequential recommendation based on a multi-interest framework aims to analyze different aspects of interest historical interactions and generate predictions user’s potential in list items. Most existing methods only focus what are the multiple interests behind but neglect evolution user over time. To explore impact temporal dynamics extraction, this paper explicitly models timestamp with network proposes time-highlighted learn preferences, which considers not at moments also possible trends...

Although Web prefetching is regarded as an effective method to improve client access performance, the associated overhead prevents it from being widely deployed. Specifically, a major weakness in existing servers that activities are scheduled independently of dynamically changing server workloads. Without proper control and coordination between two kinds activities, can negatively affect services degrade performance. We first develop open queuing model characterize detailed transactions...

To solve the problem of roll eccentricity signal with noise in HAGC system, a new method based on adaptive threshold de-noising algorithm is proposed. The can self-adaptively decide wavelet analysis and make SNR as function parameter filter to acquire optimal by using midpoint method. This has properties self-adaptation, strong robustness, low calculation, simple arithmetic so on. Simulation shows that result this much better than soft-thresholding hard-thresholding good for system.

The imaging process of SAR sea ice image is blurred by random factors, resulting in unclear image, which increases the difficulty automatic interpretation images. In view above problems, this paper proposes an classification images combined with Retinex and Gaussian Mixture Model algorithm (R-gmm). Firstly, convoluted function, then optimized EM GMM model, finally output obtained. experimental results show that effectively enhances sharpness improves segmentation accuracy Promotes...