- 标题
- 摘要
- 关键词
- 实验方案
- 产品
过滤筛选
- 2018
- Conditional Random Fields (CRF)
- Convolutional Neural Network (CNN)
- Fine Classification
- Airborne hyperspectral
- Optoelectronic Information Science and Engineering
- Wuhan University
- Central South University
- Hubei University
-
[IEEE 2018 IEEE 7th Global Conference on Consumer Electronics (GCCE) - Nara, Japan (2018.10.9-2018.10.12)] 2018 IEEE 7th Global Conference on Consumer Electronics (GCCE) - Sensorless Contact Force Evaluation of Bone-conducted Sound Transducer by Electrical Impedance in Limited Frequency Range
摘要： A contact estimation method of bone-conducted sound transducer was evaluated with data in limited frequency ranges. The estimation method uses three-layered neural network to estimate the contact force from electrical impedance. An experiment was performed with 12 subjects to evaluate contact force estimation accuracy. The electrical impedance was measured for 300 points in a frequency range of 10 Hz to 60 kHz. For evaluating the estimation accuracy with different frequency ranges, 464 combinations of start and stop frequencies were evaluated. The result showed that it is possible to reduce the frequency range without compromise on estimation error. With the limited frequency range, the measurement time can be reduced up to 5.4% of the 10 H–60 kHz range.
关键词： bone-conducted sound transducer,contact force estimation,neural network,frequency range,electrical impedance
更新于2025-09-04 15:30:14
-
[IEEE 2018 IEEE 7th Global Conference on Consumer Electronics (GCCE) - Nara, Japan (2018.10.9-2018.10.12)] 2018 IEEE 7th Global Conference on Consumer Electronics (GCCE) - Non-blind Image Restoration Based on Convolutional Neural Network
摘要： Blind image restoration processors based on convolutional neural network (CNN) are intensively researched because of their high performance. However, they are too sensitive to the perturbation of the degradation model. They easily fail to restore the image whose degradation model is slightly different from the trained degradation model. In this paper, we propose a non-blind CNN-based image restoration processor, aiming to be robust against a perturbation of the degradation model compared to the blind restoration processor. Experimental comparisons demonstrate that the proposed non-blind CNN-based image restoration processor can robustly restore images compared to existing blind CNN-based image restoration processors.
关键词： convolutional neural network,non-blind image restoration
更新于2025-09-04 15:30:14
-
[IEEE 2018 IEEE 7th Global Conference on Consumer Electronics (GCCE) - Nara, Japan (2018.10.9-2018.10.12)] 2018 IEEE 7th Global Conference on Consumer Electronics (GCCE) - Improve Visibility of Nighttime Images for Pedestrian Recognition by In-Vehicle Camera
摘要： In this research, we propose a visibility improvement method for pedestrian recognition on night vehicle camera images. In our method, we input the nighttime images and the corresponding daytime images prepared by simulation to the Neural Network, which is one of Deep Learning, and learn the feature. After that, we were able to perform image conversion experiments in the range of 5 m to 30 m from the own vehicle to the pedestrian using the model obtained by training, and to convert the image to bring the night image closer to the daytime image.
关键词： sequential images,monocular camera,pedestrian detection,Deep Learning,image convert,Neural Network
更新于2025-09-04 15:30:14
-
[IEEE 2018 International Conference on Advances in Computing, Communications and Informatics (ICACCI) - Bangalore, India (2018.9.19-2018.9.22)] 2018 International Conference on Advances in Computing, Communications and Informatics (ICACCI) - Enhanced Deep Image Super-Resolution
摘要： Recent advances in deep learning have facilitated new modalities for transforming the lower resolution image to higher resolution. The generated high resolution image must reconstruct the high frequency details of the image to generate a plausible result. To facilitate feature reuse for the task of super-resolution, we propose residual learning based convolutional neural network architecture. A pixel shuffle operation is performed in the upsampling procedure to mitigate the commonly encountered problem of artifacts in the predicted high resolution image. Our model makes use of a joint loss function consisting of pixel-wise loss and feature loss to learn the mapping from low resolution to its high resolution version. Additionally, our model has the ability to progressively increment to perform multi-scale super-resolution. An extensive experiment is performed to validate our model on the diverse ImageNet dataset. We show the effectiveness of our model through visual comparative assessment as well as quantitative comparative analysis with the state-of-the-art.
关键词： Residual block,Convolutional Neural Network,Image super-resolution
更新于2025-09-04 15:30:14
-
PET Image Reconstruction Using Deep Image Prior
摘要： Recently deep neural networks have been widely and successfully applied in computer vision tasks and attracted growing interests in medical imaging. One barrier for the application of deep neural networks to medical imaging is the need of large amounts of prior training pairs, which is not always feasible in clinical practice. This is especially true for medical image reconstruction problems, where raw data are needed. Inspired by the deep image prior framework, in this work we proposed a personalized network training method where no prior training pairs are needed, but only the patient’ own prior information. The network is updated during the iterative reconstruction process using the patient specific prior information and measured data. We formulated the maximum likelihood estimation as a constrained optimization problem and solved it using the alternating direction method of multipliers (ADMM) algorithm. Magnetic resonance imaging (MRI) guided Positron emission tomography (PET) reconstruction was employed as an example to demonstrate the effectiveness of the proposed framework. Quantification results based on simulation and real data show that the proposed reconstruction framework can outperform Gaussian post-smoothing and anatomically-guided reconstructions using the kernel method or the neural network penalty.
关键词： positron emission tomography,unsupervised learning,Medical image reconstruction,deep neural network
更新于2025-09-04 15:30:14
-
Hyperspectral Image Classification Using Spatial and Edge Features Based on Deep Learning
摘要： In recent years, deep learning has been widely used in the classification of hyperspectral images and good results have been achieved. But it is easy to ignore the edge information of the image when using the spatial features of hyperspectral images to carry out the classification experiments. In order to make full use of the advantages of convolution neural network (CNN), we extract the spatial information with the method of minimum noise fraction (MNF) and the edge information by bilateral filter. The combination of the two kinds of information not only increases the useful information but also effectively removes part of the noise. The convolution neural network is used to extract features and classify for hyperspectral images on the basis of this fused information. In addition, this article also uses another kind of edge-filtering method to amend the final classification results for a better accuracy. The proposed method was tested on three public available datasets: the University of Pavia, the Salinas, and the Indian Pines. The competitive results indicate that our approach can realize a classification of different ground targets with a very high accuracy.
关键词： hyperspectral images classification,Deep learning,spatial features,convolution neural network,minimum noise fraction
更新于2025-09-04 15:30:14
-
[IEEE 2018 25th IEEE International Conference on Image Processing (ICIP) - Athens, Greece (2018.10.7-2018.10.10)] 2018 25th IEEE International Conference on Image Processing (ICIP) - Patch-Based Stereo Matching Using 3D Convolutional Neural Networks
摘要： In this paper, we propose patch-based stereo matching using 3D convolutional neural networks (CNN). We extract spatial color and disparity features simultaneously through 3D CNN. We treat stereo matching as multi-class classification that the classes are all possible disparity values. We first generate a large set of patches from stereo images for 3D CNN. Then, we get an initial disparity map through 3D CNN and refine it using color image guided filtering. The color image guided filtering minimizes outliers and refines edges in disparity without texture copying artifacts. Experimental results show that the proposed method successfully estimates disparity in smooth and discontinuity regions while preserving edges as well as outperforms state-of-the-arts in terms of average errors.
关键词： disparity,3D convolutional neural network,Stereo matching,guided filter,patch-based
更新于2025-09-04 15:30:14
-
Switching of Wavelet Transforms by Neural Network for Image Compression
摘要： Nowadays, digital images compression requires more and more significant attention of researchers. Even when high data rates are available, image compression is necessary in order to reduce the memory used, as well the transmission cost. An ideal image compression system must yield high-quality compressed image with high compression ratio. In this article, a neural network is implemented for image compression using the feature of wavelet transform. The idea is that a back-propagation neural network can be trained to relate the image contents to its ideal compression method between two different wavelet transforms: orthogonal (Haar) and biorthogonal (bior4.4).
关键词： Scaled Conjugate Gradient,Neural Network,Image Compression,Haar Wavelet,Biorthogonal Wavelet,Back-Propagation Algorithm
更新于2025-09-04 15:30:14
-
Deep learning-based automatic volumetric damage quantification using depth camera
摘要： A depth camera or 3-dimensional scanner was used as a sensor for traditional methods to quantify the identified concrete spalling damage in terms of volume. However, to quantify the concrete spalling damage automatically, the first step is to detect (i.e., identify) the concrete spalling. The multiple spots of spalling can be possible within a single structural element or in multiple structural elements. However, there is, as of yet, no method to detect concrete spalling automatically using deep learning methods. Therefore, in this paper, a faster region-based convolutional neural network (Faster R-CNN)-based concrete spalling damage detection method is proposed with an inexpensive depth sensor to quantify multiple instances of spalling simultaneously in the same surface separately and consider multiple surfaces in structural elements. A database composed of 1091 images (with 853 × 1440 pixels) labeled for volumetric damage is developed, and the deep learning network is then modified, trained, and validated using the proposed database. The damage quantification is automatically performed by processing the depth data, identifying surfaces, and isolating the damage after merging the output from the Faster R-CNN with the depth stream of the sensor. The trained Faster R-CNN presented an average precision (AP) of 90.79%. Volume quantifications show a mean precision error (MPE) of 9.45% when considering distances from 100 cm to 250 cm between the element and the sensor. Also, an MPE of 3.24% was obtained for maximum damage depth measurements across the same distance range.
关键词： Convolutional neural network,Deep learning,Concrete spalling,Depth sensor,Volume quantification
更新于2025-09-04 15:30:14
-
Rapid prediction of acid detergent fiber content in corn stover based on NIR-spectroscopy technology
摘要： Prediction of acid detergent fiber (ADF) content in corn stover depends on precise data and appropriate analytical methods. In this paper, the optimal PLSR-BPNN model was created for rapidly getting ADF content based on the optimal selection of crucial parameters and the combination of partial least squares regression (PLSR) and back propagation neural network (BPNN). Herein, Mahalanobis distance (MD) was proposed as a tool to recognize and remove outliers. Additionally, on the basis of the characteristic bands extracted by correlation coefficient method (CC), principal component analysis (PCA) was performed to select principal components (PCs) to further compress data of bands for obtaining few characteristic wavelengths. It turned out that the performance of PLSR calibration model based on the selected 10 wavelengths was best. The correlation coefficient (R2), root mean square error of prediction (RMSEP), residual predictive deviation (RPD) and relative standard deviation (RSD) of test set successively were 0.9936, 0.3765, 12.5869, and 0.0087. Besides, BPNN was proposed to cut down the nonlinear regression residual of PLSR model. Genetic algorithm (GA) was applied to avoid the problem of local minimum in network. When RMSEP decreased to the minimum value of 0.2181, PLSR-BPNN model was proven to further improve performance and reached for the best level. Finally, the result of external validation shown that the R2, RMSEP, RPD, RSD were 0.9856, 0.4590, 8.3264, 0.0110, respectively, the created model presented the best predictive performance. Hence, the proposed methods combining with NIR-spectroscopy technology can be used to determine ADF content in corn stover.
关键词： Principal component analysis,Corn stover,Acid detergent fiber,Back propagation neural network,Genetic algorithm,Partial least squares regression
更新于2025-09-04 15:30:14