The performance of synthetic aperture radar (SAR) image change detection primarily depends on the quality measure and accuracy analysis. To these ends, this article presents a heterogeneous graph-driven method for unsupervised between SAR images. First, novel graph that incorporates both local global connections is rigorously constructed to excavate affinities from different neighbourhoods. Then, composite random walk matrix, derived graph, designed quantify subsequently utilized compose an...

A novel color image segmentation method based on finite Gaussian mixture model is proposed in this paper. First, we use EM algorithm to estimate the distribution of input data and number components automatically determined by MML criterion. Then carried out clustering each pixel into appropriate component according maximum likelihood (ML) The advantage our lies its ability less relying initialization segmenting images a totally unsupervised manner. Experimental results show that can obtain...

Deep learning is part of the field artificial intelligence. It has powerful feature extraction and capabilities. Because its various advantages, it been applied in many fields. Object detection an important technology deep learning, object based on also studied by people. With gradual improvement people's living standards, pets have gradually received love, among which pet dogs occupy majority. Different types will bring different problems. For example, large may be more aggressive cause...

Classification algorithms integrated with convolutional neural networks (CNN) display high accuracies in synthetic aperture radar (SAR) image classification. However, their consideration of spatial information is not comprehensive and effective, which causes poor performance edges complex regions. This paper proposes a Markov random field (MRF)-based algorithm for SAR classification fully considers the constraints between superpixel Firstly, initialization region labels obtained by CNN....

In this letter, a pointwise approach using stereo-graph cuts (SGC) with spatio-temporal information is proposed for synthetic aperture radar (SAR) image change detection. The SGC based on stereo-graph, which designed to connect the local maximum pixels two SAR images and can be used capture information. With support of novel energy function presented quantify implicitly measure difference between images. Therefore, detection results obtained by minimizing graph technology. Experimental real...

A unified framework for UAV-aided satellite communication network based on non-orthogonal multiple access (NOMA) is proposed. The serves as base station with the aid of a decode-forward (DF) UAV relay to communicate two ground users using NOMA protocol. Assuming that there no direct link between and due severe fading. satellite-to-UAV assumed experience Rician fading UAV-to-users are Nakagami-m In order evaluate performance system model more directly, expressions exact asymptomatic outage...

In this letter, we propose the CLN-Fusion, a novel hybrid fusion approach that leverages merits of CNNs and vision Transformers (ViTs) to couple local-nonlocal feature representations between SAR MS images. Specifically, construct paired token projection (PTP) match observation scenario content consistency two. Meanwhile, in terms merging complementary features structures images textures images, establish pyramid CNN ViT branches assemble two pure volumes with convolutional inductive biases...

Conditional random fields (CRFs), a particular type of graph neural networks (GNNs), can be used to make structured predictions in machine learning, with various applications from image processing and natural language recommender systems. CRFs refine the prediction sample by taking into account its context information. However, there is lack work on post-hoc explanation approaches CRFs, especially when model softmax-activated like deep mean field network (DMFN). In this paper, we bridge gap...

Recently, convolutional neural networks (CNNs) have been widely used for synthetic aperture radar (SAR) image classification because of their powerful feature extraction ability and high performance. However, extracting discriminative features with limited training samples is still a challenge. Moreover, some may be edge samples, which often contain multiple categories, thus deteriorate accuracy. To address these issues, we propose novel framework, named dense-connected global covariance...

To have higher resolution and precision, the amount of industrial computed tomography data has become larger larger. Moreover, images are approximately piece-wise constant, which fits for encoding contour. Then, we develop an improved compression method based on wavelet contour coding Firstly, merge Freeman idea into our IMCE (an contours extraction stationary wavelet) to extract contours. Simultaneously, each point extracted by is directly stored recording relative coordinates not actual...

It is always the top priority for network service providers to provide better Quality of Experience (QoE) users around world. Huge efforts have been devoted Adaptive Bit Rate (ABR) streaming, but it only a passive QoE amelioration method by adapting bandwidth variations. In this paper, we design Hierarchical Navigable Small World (HNSW) query algorithm and classify help content provider improve users' under amount playback logs with different network-context features. To make prediction...