- 标题
- 摘要
- 关键词
- 实验方案
- 产品
-
Superpixel-Based Semisupervised Active Learning for Hyperspectral Image Classification
摘要： In this work, we propose a new semisupervised active learning approach for hyperspectral image classification. The proposed method aims at improving machine generalization by using pseudolabeled samples, both confident and informative, which are automatically and actively selected, via semisupervised learning. The learning is performed under two assumptions: a local one for the labeling via a superpixel-based constraint dedicated to the spatial homogeneity and adaptivity into the pseudolabels, and a global one modeling the data density by a multinomial logistic regressor with a Markov random field regularizer. Furthermore, we propose a density-peak-based augmentation strategy for pseudolabels, due to the fact that the samples without manual labels in their superpixel neighborhoods are out of reach for the automatic sampling. Three real hyperspectral datasets were used in our experiments to evaluate the effectiveness of the proposed superpixel-based semisupervised learning approach. The obtained results indicate that the proposed approach can greatly improve the potential for semisupervised learning in hyperspectral image classification.
关键词： semisupervised learning,hyperspectral image classification,superpixel,clustering,Active learning
更新于2025-09-23 15:23:52
-
A change detection framework by fusing threshold and clustering methods for optical medium resolution remote sensing images
摘要： In change detection (CD) of medium-resolution remote sensing images, the threshold and clustering methods are two kinds of the most popular ones. It is found that the threshold method of the expectation maximization (EM) algorithm usually generates a CD map including many false alarms but almost detecting all changes, and the fuzzy local information c-means algorithm (FLICM) obtains a homogenous CD map but with some missed detections. Therefore, a framework is designed to improve CD results by fusing the advantages of the threshold and clustering methods. The CD map generated by the clustering method of FLICM is used to remove false alarms in the CD map obtained by EM threshold method by an overlap fusion. Then, the local Markov random field model is implemented to verify the potentially missed detections. Finally, a fused CD map with less false alarms and missed detections is achieved. Two experiments were carried out on two Landsat ETM+ data sets. The proposed method obtained the least errors (1.11% and 3.51%) and the highest kappa coefficient (0.9366 and 0.8834), respectively, when compared with five popular CD methods.
关键词： Change detection,advantage fusion,remote sensing,clustering,threshold
更新于2025-09-23 15:23:52
-
Analysis of trajectory similarity and configuration similarity in on-the-fly surface-hopping simulation on multi-channel nonadiabatic photoisomerization dynamics
摘要： We propose an “automatic” approach to analyze the results of the on-the-fly trajectory surface hopping simulation on the multi-channel nonadiabatic photoisomerization dynamics by considering the trajectory similarity and the configuration similarity. We choose a representative system phytochromobilin (PΦB) chromophore model to illustrate the analysis protocol. After a large number of trajectories are obtained, it is possible to define the similarity of different trajectories by the Fréchet distance and to employ the trajectory clustering analysis to divide all trajectories into several clusters. Each cluster in principle represents a photoinduced isomerization reaction channel. This idea provides an effective approach to understand the branching ratio of the multi-channel photoisomerization dynamics. For each cluster, the dimensionality reduction is employed to understand the configuration similarity in the trajectory propagation, which provides the understanding of the major geometry evolution features in each reaction channel. The results show that this analysis protocol not only assigns all trajectories into different photoisomerization reaction channels but also extracts the major molecular motion without the requirement of the pre-known knowledge of the active photoisomerization site. As a side product of this analysis tool, it is also easy to find the so-called “typical” or “representative” trajectory for each reaction channel.
关键词： trajectory similarity,multi-channel nonadiabatic photoisomerization dynamics,Fréchet distance,dimensionality reduction,phytochromobilin chromophore,on-the-fly surface-hopping simulation,configuration similarity,clustering analysis
更新于2025-09-23 15:23:52
-
[IEEE IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Valencia (2018.7.22-2018.7.27)] IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Sea Ice Change Detection in SAR Images Based on Collaborative Representation
摘要： Sea ice change detection from synthetic aperture radar (SAR) images is important for navigation safety and natural resource extraction. This paper proposed a sea ice change detection method from SAR images based on collaborative representation. First, neighborhood-based ratio is used to generate a difference image (DI). Then, some reliable samples are selected from the DI by hierarchical fuzzy C-means (FCM) clustering. Finally, based upon these samples, collaborative representation method is utilized to classify pixels from the original SAR images into unchanged and changed class. From there, the final change map can be obtained. Experimental results on two real sea ice datasets demonstrate the superiority of the proposed method over two closely related methods.
关键词： sea ice change detection,synthetic aperture radar,clustering method,collaborative representation
更新于2025-09-23 15:23:52
-
[IEEE 2018 10th IAPR Workshop on Pattern Recognition in Remote Sensing (PRRS) - Beijing (2018.8.19-2018.8.20)] 2018 10th IAPR Workshop on Pattern Recognition in Remote Sensing (PRRS) - FPGA Optimization for Hyperspectral Target Detection with Collaborative Representation
摘要： Most applications based on wireless sensor networks (WSN) have devices with constraints of limited energy and computational/storage capabilities. The traditional security mechanisms are not desirable to these applications. A lightweight security and energy-efficient clustering protocol was proposed in this paper to solve the security problem in the clustering-based sensor networks. Firstly, a lightweight security algorithm is proposed to meet the security requirements, which reduces the communication overload by using the transmission key index. Secondly, in the process of clustering, the base station (BS) and cluster head (CH) use lightweight authentication procedure to verify the identities hierarchically, to reduce the risk of attacks from malicious nodes posing as BS or CH. Thirdly, the proposed protocol is analyzed in the aspects of security and energy consumption. Simulation results show that the proposed protocol not only enhances the network security but also improves the energy efficiency.
关键词： Lightweight security,Energy-efficient protocol,Clustering WSN
更新于2025-09-23 15:23:52
-
Segmented and non-segmented stacked denoising autoencoder for hyperspectral band reduction
摘要： Hyperspectral image (HSI) analysis often requires selecting the most informative bands instead of processing the whole data without losing the key information. Existing band reduction (BR) methods have the capability to reveal the nonlinear properties exhibited in the data but at the expense of losing its original representation. To cope with the said issue, an unsupervised non-linear segmented and non-segmented stacked denoising autoencoder (UDAE)-based BR method is proposed. Our aim is to find an optimal mapping and construct a lower-dimensional space that has a similar structure to the original data with least reconstruction error. The proposed method first confronts the original HS data into smaller regions in the spatial domain and then each region is processed by UDAE individually. This results in reduced complexity and improved efficiency of BR for classification. Our experiments on publicly available HS datasets with various types of classifiers demonstrate the effectiveness of UDAE method which equates favorably with other state-of-the-art dimensionality reduction and BR methods.
关键词： Autoencoder (AE),Hyperspectral imaging (HSI),Classification,Clustering,Band reduction (BR)
更新于2025-09-23 15:23:52
-
[IEEE 2018 IEEE International Instrumentation and Measurement Technology Conference (I2MTC) - Houston, TX (2018.5.14-2018.5.17)] 2018 IEEE International Instrumentation and Measurement Technology Conference (I2MTC) - Multi-core cable fault diagnosis using cluster time-frequency domain reflectometry
摘要： Guaranteeing the integrity and functionality of the control and instrumentation (C&I) cable system is essential in ensuring safe nuclear power plant (NPP) operation. When a fault occurs in a multi-core cable, it not only affects the signals of faulty lines but in fact, disturbs the rest as well due to crosstalk and noise interference. Therefore, this results in C&I signal errors in NPP operation and further leads to a rise in concern regarding the NPP operation. Thus, it is necessary for diagnostic technologies of multi-core C&I cables to classify the faulty line and detect the fault to assure the safety and reliability of NPP operation. We propose a diagnostic method that detects the fault location and faulty line in multi-core C&I cable using a clustering algorithm based on TFDR results. The faulty line detection clustering algorithm uses TFDR cross-correlation and phase synchrony results as input feature data altogether which can detect the faulty line and identify the fault point successfully. The proposed clustering algorithm is verified by experiments with two possible fault scenarios in NPP operation.
关键词： fault diagnosis,reflectometry,control and instrumentation cable,K-means clustering,crosstalk,time-frequency analysis
更新于2025-09-23 15:23:52
-
[IEEE 2018 Eighth International Conference on Image Processing Theory, Tools and Applications (IPTA) - Xi'an, China (2018.11.7-2018.11.10)] 2018 Eighth International Conference on Image Processing Theory, Tools and Applications (IPTA) - Joint Deep Learning and Clustering Algorithm for Liquid Particle Detection of Pharmaceutical Injection
摘要： At present, the detection of pharmaceutical injection products is a quite important step in the pharmaceutical manufacturing, as it has the direct related to the quality of medical product quality. Aiming at the difficulty that liquid particle has a smaller pixel point in the high resolution image of detection of pharmaceutical liquid particle, hence consider combined with deep neural network and clustering algorithm for detection and localization of little particle, and a processing method combining single frame images with multi-frame images was proposed to identifying liquid particle. Firstly, the single-frame image is detected by using Faster-RCNN deep neural network, and it can obtain the detection result of the 8-frame sequence image. Then hierarchical clustering and K-means clustering algorithm are used for clustering to obtain the same target motion area. In this way, liquid particle can be more accurately identified and the accuracy of detection can be greatly improved. The experimental results show that the accuracy of detection and recognition of foreign substances in liquid medicine is improved by more than 10% on average.
关键词： Liquid particle detection,Injection detection,K-means clustering,Hierarchical clustering,Faster-RCNN
更新于2025-09-23 15:22:29
-
Task-Oriented GAN for PolSAR Image Classification and Clustering
摘要： Based on a generative adversarial network (GAN), a novel version named Task-Oriented GAN is proposed to tackle difficulties in PolSAR image interpretation, including PolSAR data analysis and small sample problem. Besides two typical parts in GAN, i.e., generator (G-Net) and discriminator (D-Net), there is a third part named TaskNet (T-Net) in the Task-Oriented GAN, where T-Net is employed to accomplish a certain task. Two tasks, PolSAR image classification and clustering, are studied in this paper, where T-Net acts as a Classifier and a Clusterer, respectively. The learning procedure of Task-Oriented GAN consists of two main stages. In the first stage, G-Net and D-Net vie with each other like that in a general GAN; in the second stage, G-Net is adjusted and oriented by T-Net so that more samples, which are benefit for the task and called fake data, are generated. As a result, Task-Oriented GAN not only has the advantage of GAN (no-assumption data modeling) but also overcomes the disadvantage of GAN (task-free). After learning, fake data are employed to enrich training set and avoid overfitting; so Task-Oriented GAN performs well even if the manual-labeled data are small. To verify the effectiveness of T-Net, a visualized comparison is provided, where some fake digits generated from Task-Oriented GAN are illustrated along with that from GAN. What is more, considering that there is a great difference between PolSAR data and general data, in our PolSAR image classification and clustering tasks, the specific PolSAR information is inserted into the structure of the Task-Oriented GAN. This enables researchers to mine inherent information in PolSAR data without any data hypothesis and find ways for small sample problem at the same time. Experiment results tested on three PolSAR images show that the proposed method performs well in dealing with PolSAR image classification and clustering.
关键词： generative adversarial network (GAN),task-oriented,Clustering,PolSAR image classification
更新于2025-09-23 15:22:29
-
[IEEE 2018 XIV International Scientific-Technical Conference on Actual Problems of Electronics Instrument Engineering (APEIE) - Novosibirsk, Russia (2018.10.2-2018.10.6)] 2018 XIV International Scientific-Technical Conference on Actual Problems of Electronics Instrument Engineering (APEIE) - Increase in Accuracy of the Solution of the Problem of Identification of Production Batches of Semiconductor Devices
摘要： Authors provide a comparative analysis of the results of applying various automatic grouping (clustering) algorithms to a problem of separating homogeneous batches of electrical radio products from a mixed batch. Authors consider opportunities of optimizing the algorithm parameters as well as compilation of ensembles of clustering algorithms. The obtained results demonstrate that the homogeneity analysis of batches of the electronic components installed in real space systems is an actual problem, especially in the case of A high level of integration. Authors substantiate the necessity of centralized acquisition of spacecraft equipment on the principle of equal reliability by creating special production batches manufactured for space industry.
关键词： clustering algorithms,space electronics,ensembles of algorithms
更新于2025-09-23 15:22:29