In this paper, we present an improved fuzzy C-means (FCM) algorithm for image segmentation by introducing a tradeoff weighted factor and kernel metric. The depends on the space distance of all neighboring pixels their gray-level difference simultaneously. By using factor, new can accurately estimate damping extent pixels. order to further enhance its robustness noise outliers, introduce measure objective function. adaptively determines parameter fast bandwidth selection rule based variance...

Remote sensing (RS) scene classification (RSSC) is a prominent research topic in the RS community. Multilevel feature fusion an important way of addressing classification, and many methods have been proposed recent years. Although they succeed, current can still be improved, particularly distinguishing contributions different multilevel features fully effectively fusing them. To address above issues exploit potential for tasks, we propose new model named multiscale sparse cross-attention...

This letter presents a novel method based on wavelet fusion for change detection in synthetic aperture radar (SAR) images. The proposed approach is applied to generate the difference image (DI) by using complementary information from mean-ratio and log-ratio To restrain background (unchanged areas) enhance of changed regions fused DI, rules weight averaging minimum standard deviation are chosen fuse coefficients low- high-frequency bands, respectively. Experiments real SAR images confirm...

The content-based remote sensing image retrieval (CBRSIR) has attracted increasing attention with the number of (RS) images growing explosively. Benefiting from strong capacity deep convolutional neural network (DCNN), performance CBRSIR been improved in recent years. Although great successes have obtained, learning RS images' representative features and enhancing efficiency for large-scale tasks are still two challenging problems. In this article, we propose a new method named feature hash...

Due to the lack of label information and intrinsic complexity hyperspectral images (HSIs), unsupervised band selection is always one most challenging tasks in HSI processing. Fuzzy clustering a promising technique for selection, which can partition unlabeled data into groups effectively. However, due limits its optimization process, standard fuzzy sensitive initialization easy be trapped local optimum. To address limits, novel method proposed, combining with particle swarm (PSO). A newly...

Cross-modal remote sensing image-text retrieval (CMRSITR) is a challenging topic in the (RS) community. It has gained growing attention because it can be flexibly used many practical applications. In current deep era, with help of convolutional neural networks (DCNNs), successful CMRSITR methods have been proposed. Most them first learn valuable features from RS images and texts respectively. Then, obtained visual textual are mapped into common space for final retrieval. The above operations...

change of blood glucose (BG) level stimulates the autonomic nervous system leading to variation in both human's electrocardiogram (ECG) and photoplethysmogram (PPG). In this article, we aimed construct a novel multimodal framework based on ECG PPG signal fusion establish universal BG monitoring model. This is proposed as spatiotemporal decision strategy that uses weight-based Choquet integral for monitoring. Specifically, performs three-level fusion. First, signals are collected coupled into...

There is an increasing recognition on community detection in complex networks recent years. In this study, we improve a recently proposed memetic algorithm for networks. By introducing Population Generation via Label Propagation (PGLP) tactic, Elitism Strategy (ES) and Improved Simulated Annealing Combined Local Search (ISACLS) strategy, the improved called (iMeme-Net) put forward solving problems. Experiments both computer-generated real-world show effectiveness multi-resolution ability of method.

With the explosive growth of volume and resolution high-resolution remote-sensing (HRRS) images, management them becomes a challenging task. The traditional content-based image retrieval (CBRSIR) technologies cannot meet what we expect due to large archives complex contents within HRRS images. As successful approximate nearest neighborhood (ANN) search technique, Hash learning has received wide attention, especially when deep convolutional neural networks (DCNNs) appear. Due DCNNs’...

Remote sensing image retrieval (RSIR), a superior content organization technique, plays an important role in the remote (RS) community. With number of RS images increases explosively, not only precision but also efficiency is emphasized large-scale RSIR scenario. Therefore, approximate nearest neighborhood (ANN) search attracts researchers’ attention increasingly. In this paper, we propose new hash learning method, named semi-supervised deep adversarial hashing (SDAH), to accomplish ANN for...

In this paper, we propose a novel unsupervised evolutionary clustering algorithm for mixed type data, k-prototype (EKP). As partitional algorithm, (KP) is well-known one data. However, it sensitive to initialization and converges local optimum easily. Global searching ability of the most important advantages (EA), so an EA framework introduced help KP overcome its flaws. study, applied as search strategy, runs under control framework. Experiments on synthetic real-life datasets show that EKP...

This paper presents the first characterization and modeling of dynamic propagation channels for human body communication (HBC). In-situ experiments were performed using customized transceivers in an anechoic chamber. Three HBC channels, i.e., from right leg to left leg, hand investigated under thirty-three motion scenarios. Snapshots data (2,800,000) acquired five volunteers. Various path gains caused by different locations movements quantified statistical distributions estimated. In...

This letter investigated characteristics of human body as a communication channel at 45 MHz, in different surrounding environments, taking into account parameter changes arising from activities. A large amount measurement data has been collected five real environments for random motions: conference hall, laboratory, grove, playground, and an anechoic chamber. The received power nine (HBC) channels was acquired, with more than 2 700 000 snapshots total. Environmentally independent analysis...

Community structure is one of the most important properties in social networks. In dynamic networks, there are two conflicting criteria that need to be considered. One snapshot quality, which evaluates quality community partitions at current time step. The other temporal cost, difference between communities different steps. this paper, we propose a decomposition-based multiobjective detection algorithm simultaneously optimize these objectives reveal and its evolution It employs framework...

Here, the authors propose a novel two-phase clustering algorithm with density exploring distance (DED) measure. In first phase, fast global K -means is used to obtain cluster number and prototypes. Then, prototypes of all these clusters representatives points belonging are regarded as input data set second phase. Afterwards, clustered according DED measure which makes locating in same structure possess high similarity each other. experimental studies, test proposed on seven artificial well...

In the biomedical field, large amounts of biological and clinical data have been accumulated rapidly, which can be analyzed to emphasize assessment at-risk patients improve diagnosis. However, a major challenge encountered associated with analysis is so-called “curse dimensionality.” For this issue, novel feature selection method based on an improved binary clonal flower pollination algorithm proposed eliminate unnecessary features ensure highly accurate classification disease. The absolute...

Real-world optimization involving multiple objectives in changing environment known as dynamic multi-objective (DMO) is a challenging task, especially special regions are preferred by decision maker (DM). Based on novel preference dominance concept called sphere-dominance and the theory of artificial immune system (AIS), immune-inspired algorithm (SPIA) proposed for DMO this paper. The main contributions SPIA its mechanism sampling study, which based probability statistics, respectively....