检索结果-内蒙古大学图书馆

AN ANISOTROPIC SCATTERING ANALYSIS METHOD BASED ON LIKELIHOOD RATIO USING CIRCULAR SAR DATA

学校读者我要写书评

暂无评论

AN ANISOTROPIC SCATTERING ANALYSIS METHOD BASED ON LIKELIHOO...

IEEE International Geoscience and Remote Sensing Symposium

作者： Fei Teng Wen Hong Yun Lin Bing Han Yanping Wang Wenjie Shen Shanshan Feng Key Laboratory of Technology in Geospatial Information Processing and Application System Chinese Academy of Sciences Beijing China North China University of Technology Beijing China

The scattering of an anisotropic target is aspect dependent. Circular SAR (CSAR) can observe the scattering behavior in different aspect angles. In this paper, we propose an anisotropy scattering analysis method based on the likelihood ratio using CSAR data. CSAR data is used to provide sub-aperture images in different aspect angles. The likelihood ratio is defined as the ratio of the conditional probability under two hypotheses, anisotropic and isotropic. Anisotropic and isotropic scatterings can be discriminated by the value of the likelihood ratio. The scattering direction of the anisotropic scattering can be obtained by using our method too. We use a C-band CSAR data, which is acquired by the Institute of Electronics, Chinese Academy of Sciences (IEcas) to validate our method.

关键词： Apertures Image color analysis Radar polarimetry Trajectory Optical imaging Optical scattering

Stereoscopic omnidirectional image quality assessment based on predictive coding theory

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Chen, Zhibo Xu, Jiahua Lin, Chaoyi Zhou, Wei CAS Key Laboratory of Technology in Geo-Spatial Information Processing and Application System University of Science and Technology of China Hefei Anhui230027 China

Objective quality assessment of stereoscopic omnidirectional images is a challenging problem since it is influenced by multiple aspects such as projection deformation, field of view (FoV) range, binocular vision, visual comfort, etc. Existing studies show that classic 2D or 3D image quality assessment (IQA) metrics are not able to perform well for stereoscopic omnidirectional images. However, very few research works have focused on evaluating the perceptual visual quality of omnidirectional images, especially for stereoscopic omnidirectional images. In this paper, based on the predictive coding theory of the human vision system (HVS), we propose a stereoscopic omnidirectional image quality evaluator (SOIQE) to cope with the characteristics of 3D 360-degree images. Two modules are involved in SOIQE: predictive coding theory based binocular rivalry module and multi-view fusion module. In the binocular rivalry module, we introduce predictive coding theory to simulate the competition between high-level patterns and calculate the similarity and rivalry dominance to obtain the quality scores of viewport images. Moreover, we develop the multi-view fusion module to aggregate the quality scores of viewport images with the help of both content weight and location weight. The proposed SOIQE is a parametric model without necessary of regression learning, which ensures its interpretability and generalization performance. Experimental results on our published stereoscopic omnidirectional image quality assessment database (SOLID) demonstrate that our proposed SOIQE method outperforms state-of-the-art metrics. Furthermore, we also verify the effectiveness of each proposed module on both public stereoscopic image datasets and panoramic image datasets. Copyright © 2019, The Authors. All rights reserved.

关键词： Binocular vision

Online filter clustering and pruning for efficient convnets

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Zhou, Zhengguang Zhou, Wengang Hong, Richang Li, Houqiang CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System EEIS Department University of Science and Technology of China HeFei University of Technology

Pruning filters is an effective method for accelerating deep neural networks (DNNs), but most existing approaches prune filters on a pre-trained network directly which limits in acceleration. Although each filter has its own effect in DNNs, but if two filters are same with each other, we could prune one safely. In this paper, we add an extra cluster loss term in the loss function which can force filters in each cluster to be similar online. After training, we keep one filter in each cluster and prune others and fine-tune the pruned network to compensate the loss. Particularly, the clusters in every layer can be defined firstly which is effective for pruning DNNs within residual blocks. Extensive experiments on CIFAR10 and CIFAR100 benchmarks demonstrate the competitive performance of our proposed filter pruning method. Copyright © 2019, The Authors. All rights reserved.

关键词： Deep neural networks

Self-Tuning Asynchronous Filter for Linear Gaussian system and applications

学校读者我要写书评

暂无评论

IEEE/CAA Journal of Automatica Sinica 2018年第6期5卷 1054-1061页

作者： Wenjun Lv Yu Kang Yunbo Zhao Department of Automation University of Science and Technology of ChinaHefei 230027China State Key Laboratory of Fire Science Department of Automation and Institute of Advanced TechnologyUniversity of Science and Technology of ChinaHefei 230027China Key Laboratory of Technology in GeoSpatial Information Processing and Application System Chinese Academy of SciencesBeijing 100190China College of Information Engineering Zhejiang University of TechnologyHangzhou 310023China

In this paper, optimal filtering problem for a class of linear Gaussian systems is studied. The system states are updated at a fast uniform sampling rate and the measurements are sampled at a slow uniform sampling rate. The updating rate of system states is several times the sampling rate of measurements and the multiple is constant. To solve the problem,we will propose a self-tuning asynchronous filter whose contributions are twofold. First, the optimal filter at the sampling times when the measurements are available is derived in the linear minimum variance sense. Furthermore, considering the variation of noise statistics, a regulator is introduced to adjust the filtering coefficients adaptively. The case studies of wheeled robot navigation system and air quality evaluation system will show the effectiveness and practicability in engineering.

关键词： Air quality evaluation system linear Gaussian system wheeled robot navigation system

Image-to-image translation with multi-path consistency regularization

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Lin, Jianxin Xia, Yingce Wang, Yijun Qin, Tao Chen, Zhibo CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System University of Science and Technology of China University of Science and Technology of China Microsoft Research Asia

Image translation across different domains has attracted much attention in both machine learning and computer vision communities. Taking the translation from source domain Ds to target domain Dt as an example, existing algorithms mainly rely on two kinds of loss for training: One is the discrimination loss, which is used to differentiate images generated by the models and natural images;the other is the reconstruction loss, which measures the difference between an original image and the reconstructed version through Ds → Dt → Ds translation. In this work, we introduce a new kind of loss, multi-path consistency loss, which evaluates the differences between direct translation Ds → Dt and indirect translation Ds → Da → Dt with Da as an auxiliary domain, to regularize training. For multi-domain translation (at least, three) which focuses on building translation models between any two domains, at each training iteration, we randomly select three domains, set them respectively as the source, auxiliary and target domains, build the multi-path consistency loss and optimize the network. For two-domain translation, we need to introduce an additional auxiliary domain and construct the multi-path consistency loss. We conduct various experiments to demonstrate the effectiveness of our proposed methods, including face-toface translation, paint-to-photo translation, and deraining/de-noising translation. Copyright © 2019, The Authors. All rights reserved.

关键词： Iterative methods

LUneXt: Simple and Efficient U-shaped Network Design for Medical Image Segmentation with Nonlinear Activation

学校读者我要写书评

暂无评论

Computing Open

Computing Open 2024年第2期computing-open卷

作者： Guanghong Deng Bing Yu Wenlong Jing Yong Li Xiaodan Zhao GuangDong Engineering Technology Research Center of UAV Remote Sensing Network Guangzhou iMapCloud Intelligent Technology Co. Ltd. Guangzhou 510095 P. R. China Guangdong Province Engineering Laboratory for Geographic Spatiotemporal Big Data Key Laboratory of Guangdong for Utilization of Remote Sensing and Geographical Information System Guangdong Open Laboratory of Geospatial Information Technology and Application Guangzhou Institute of Geography Guangdong Academy of Sciences Guangzhou 510070 P. R. China Foshan iMapCloud Intelligent Technology Co. Ltd. Foshan 528251 P. R. China

Medical image segmentation has always been a challenging task. This paper proposes a new LUneXt medical image segmentation model based on the characteristics analysis of medical image data sets and testing of different nonlinear activation units. Both normalized activations for the original negative input image activation have good optimization capabilities for the tokenization module parameters proposed in the original UneXt model. Using different activation coefficients for different foreground and background areas has achieved better results. The experimental results of this paper on the Breast Ultrasound Images (BUSI) data set reached an intersection over union (IoU) value of 62.64%, a Dice value of 76.12%, and a single inference speed of 807.57 ms. The experimental IoU value of the International Skin Imaging Collaboration (ISIC 2018) data set reached 82.95%, and the Dice value reached 90.50%. The single inference speed reached 842.58 ms. The LUneXt model is more robust than other models. While improving model performance, it does not introduce higher computational complexity and does not have a major impact on the processing speed of a single image.

关键词： Image segmentation medical images neural networks activation function

Design of Command and Dispatch system for Automatic Reading of Meter Images in Substations by UAV Inspection Photos

学校读者我要写书评

暂无评论

Computing Open

Computing Open 2024年第2期computing-open卷

作者： Guanghong Deng Liangpei Lin Baihao Lin Wenlong Jing Xiaodan Zhao Yong Li Rui Yang Shidun Xie Pengfeng Wei GuangDong Engineering Technology Research Center of UAV Remote Sensing Network Guangzhou iMapCloud Intelligent Technology Co. Ltd. Guangzhou 510095 P. R. China Guangdong Province Engineering Laboratory for Geographic Spatiotemporal Big Data Key Laboratory of Guangdong for Utilization of Remote Sensing and Geographical Information System Guangdong Open Laboratory of Geospatial Information Technology and Application Guangzhou Institute of Geography Guangdong Academy of Sciences Guangzhou 510070 P. R. China Foshan iMapCloud Intelligent Technology Co. Ltd. Foshan 528251 P. R. China

The application of the combination of unmanned aerial vehicles (UAVs) and artificial intelligence is a hot topic in the intelligent inspection of substations, and meter reading is a very challenging task. This paper proposes a method based on the combination of YOLOv6n object detection and Deeplabv3 + image segmentation and performs post-processing on the segmented images to obtain meter readings. First, YOLOv6n is used to detect the meter area of the aerial image and classify the meters. Second, the detected meter images are fed into the image segmentation model. The backbone network of the Deeplabv3 + algorithm is improved by using the MobileNetv3 network, which not only effectively extracts pointers and scales, but also makes the model more lightweight. Third, License Plate Recognition Network (LPRNet) is used to recognize digital meter images. In order to solve the problem of inaccurate pointer meter readings, to begin with, the segmented image is corroded; in addition, the circular dial area is flattened into a rectangular area by concentric circle sampling method. Finally, the meter reading is calculated by the position of the pointer, the scale and the total range of the meter. The post-processing part uses numba to optimize the inference speed. The experimental results show that in two datasets, The mean average precision of 50 (mAP50) accuracy of the YOLOv6n model using this method reached 99.71% and 98.60%, respectively, and the inference speed of a single image was 17.1 ms and 13.2 ms, respectively. The mean intersection over union (mIoU) of the image segmentation model reached 82.00%, 74.73%, 73.50%, 82.26% and 73.20%, respectively, and the single segmentation speed reached 33.7 ms. The LPRNet model has a recognition accuracy of 99.17% and a single image inference speed of 14.7 ms. At the same time, several mainstream object detection and semantic segmentation algorithms are compared. The experimental results show that the method in this paper greatly im

关键词： Object detection image segmentation YOLOv6n Deeplabv3 + meter reading

Residual objectness for imbalance reduction

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Chen, Joya Liu, Dong Luo, Bin Peng, Xuezheng Xu, Tong Chen, Enhong Anhui Province Key Laboratory of Big Data Analysis and Application University of Science and Technology of China CAS Key Laboratory of Technology in Geo-Spatial Information Processing and Application System University of Science and Technology of China Tencent TaiQ Team

For a long time, object detectors have suffered from extreme imbalance between foregrounds and backgrounds. While several sampling/reweighting schemes have been explored to alleviate the imbalance, they are usually heuristic and demand laborious hyper-parameters tuning, which is hard to achieve the optimality. In this paper, we first reveal that such the imbalance could be addressed in a learning-based manner. Guided by this illuminating observation, we propose a novel Residual Objectness (ResObj) mechanism that addresses the imbalance by end-to-end optimization, while no further hand-crafted sampling/reweighting is required. Specifically, by applying multiple cascaded objectness-related modules with residual connections, we formulate an elegant consecutive refinement procedure for distinguishing the foregrounds from backgrounds, thereby progressively addressing the imbalance. Extensive experiments present the effectiveness of our method, as well as its compatibility and adaptivity for both region-based and one-stage detectors, namely, the RetinaNet-ResObj, YOLOv3-ResObj and FasterRCNN-ResObj achieve relative 3.6%, 3.9%, 3.2% Average Precision (AP) improvements compared with their vanilla models on COCO, respectively. Copyright © 2019, The Authors. All rights reserved.

关键词： Object detection

Dem Extraction Using C-Band Circular Sar Data

学校读者我要写书评

暂无评论

Dem Extraction Using C-Band Circular Sar Data

IEEE International Symposium on Geoscience and Remote Sensing (IGARSS)

作者： Shanshan Feng Yun Lin Wen Hong Bing Han Yanping Wang Yanhui Yang Wenjie Shen Fei Teng Key Laboratory of Technology in Geospatial Information Processing and Application System Chinese Academy of Sciences Beijing China North China University of Technology Beijing China Beijing Institute of Remote Sensing Equipment Beijing China

Circular Synthetic Aperture Radar(CSAR) has become a hotspot with its characteristic of elevation plane resolution and all-aspect observing ability. Digital elevation model (DEM) extraction in urban arears by using single-pass CSAR data without requiring additional knowledge is a subject of interest. The target, whose real height is not equal to the reference imaging height will project to different locations after imaging in different sub-aperture. In this paper, the quantitative relationship between offset of imaging points and height difference is deduced theoretically in the real scene, where the airborne SAR platform trajectory is not a standard circle. DEM of an area is presented using the data acquired by the Institute of Electronics, Chinese Academy of Sciences (IEcas). Compared with the DEM provided by the German Aerospace Center (DLR) with 1m absolute height error, the effectiveness of the proposed method is verified.

关键词： Synthetic aperture radar Imaging Azimuth Trajectory Radar polarimetry Data mining Geometry