Defocus blur detection (DBD) is the separation of in-focus and out-of-focus regions in an image. This process has been paid considerable attention because its remarkable potential applications. Accurate differentiation homogeneous low-contrast focal regions, as well suppression background clutter, are challenges associated with DBD. To address these issues, we propose a multi-stream bottom-top-bottom fully convolutional network (BTBNet), which first attempt to develop end-to-end deep for...

Defocus blur detection (DBD) is aimed to estimate the probability of each pixel being in-focus or out-of-focus. This process has been paid considerable attention due its remarkable potential applications. Accurate differentiation homogeneous regions and low-contrast focal regions, as well suppression background clutter, are challenges associated with DBD. To address these issues, we propose a multi-stream bottom-top-bottom fully convolutional network (BTBNet), which first attempt develop an...

Enterprise resource planning systems have required significant upgrades in the 21st century as many of obtained prior to 2000 become outdated due vendor changes. SAP and Oracle emerged dominant vendors, has announced discontinuance support future for its primary R/3 system. This study reports interviews with chief information officers (CIOs) 15 institutions that undergone (or are undergoing) enterprise system upgrades, focus on discussion major critical success factors ERP upgrade projects.

A universal fusion framework for handling multi-realm image reduces the cost of manual selection in varied applications. Addressing generality multiple realms and sensitivity specific realm, we propose a novel through learning realm-specific realm-general feature representations. Shared principle network, adaptive realm extraction strategy activation mechanism are designed facilitating high generalization across-realm specific-realm simultaneously. In addition, present no-reference...

Defocus blur detection (DBD) for natural images is a challenging vision task especially in the presence of homogeneous regions and gradual boundaries. In this paper, we propose novel image-scale-symmetric cooperative network (IS2CNet) DBD. On one hand, process image scales from large to small, IS2CNet gradually spreads recept content. Thus, region map can be optimized gradually. other small large, feels high-resolution content, thereby refining transition detection. addition, hierarchical...

Abstract With a large number of data stories expressed in diverse genres narrative visualization, storytelling is currently growing popularity the field visualization. In this paper, we investigate literature over last 10 years and suggest brand-new classification scheme for authoring tools from perspectives. Our comprehensively meticulously summarizes collected papers. By arranging papers each category by publication date, identify popular topics, present current status future directions...

Human vision is sensitive to the changes of local image details, which are actually gradients. To enhance faint infrared this article proposes a gradient field specification algorithm. First we define and histogram. Then, by analyzing characteristics histogram, construct Gaussian function obtain histogram therefore transform field. In addition, subhistogram equalization proposed based on improve contrast images. The experimental results show that algorithm can effectively weak details edges....

Infrared image segmentation is a challenging topic since infrared images are characterized by high noise, low contrast, and weak edges. Active contour models, especially gradient vector flow (GVF), have better performance for clear images. However, the GVF model has drawbacks of sensitivity to noise adaptability parameters, decreasing effect significantly. To address these problems, this paper proposes guide filter-based module (GFGVF). First, filter exploited construct novel edge map,...

Nowadays, data is being generated, collected, and analyzed at an unprecedented scale, integration the problem of combining from heterogeneous, autonomous sources, providing users with a unified view integrated data. To design framework, we need to address challenges, such as schema mapping, cleaning, record linkage, fusion. In this paper, briefly introduce traditional approaches, then, novel graph-based framework based on concept model (UCM) proposed real-world refueling problems. Within...

Computed tomography (CT) has an excellent performance in detecting dense structure, such as bones and implants, while magnetic resonance (MR) provides high-resolution information for soft issues. To obtain sufficient accurate diagnosis, we propose a CT MR image fusion method via adaptive structure decomposition to combine the complementary information. First, on basis of different scales issues, adaptively decompose source images into sub-bands (bands small, middle, large issues) by spectral...

System error registration is an important content of space-time registration, which the premise fusion multi-source measurement data. Aiming at complex electromagnetic environment interference, unrobustness traditional method was verified. A real-time median estimating algorithm slowly varying sequence based on sorting doubly linked list designed. On basis which, a robust system detection and estimation put forward. The proposed in this paper, application time-varying processing also...

In data integration, entity resolution is an important technique to improve quality. Existing researches typically assume that the target dataset only contain string-type and use single similarity metric. For larger high-dimensional dataset, redundant information needs be verified using traditional blocking or windowing techniques. this work, we propose a novel ER-resolving method hybrid approach, including type-based multiblocks, varying window size, more flexible metrics. our new ER...

Deep learning-based methods for Time Series Classification (TSC) typically utilize deep networks to extract features, which are then processed through a combination of Fully Connected (FC) layer and SoftMax function. However, we have observed the phenomenon inter-class similarity intra-class inconsistency in datasets from UCR archive further analyzed how this adversely affects "FC+SoftMax" paradigm. To address issue, introduce ECR, which, first time our knowledge, applies retrieval algorithm...

The vehicle scheduling in the logistics system of iron and steel plant is a key issue to resource allocation inventory running efficiency downstream production line. traditional strategy does not analyze cost from aspects storage order time, resulting low high cost. Taking right time arrival into considering multi-vehicle can effectively improve decrease So, paper established an integral programming model that minimize factory minimum operating conveyor belt. Then improved Glowworm Swarm...