- 标题
- 摘要
- 关键词
- 实验方案
- 产品
-
[IEEE 2018 9th International Conference on Computing, Communication and Networking Technologies (ICCCNT) - Bangalore (2018.7.10-2018.7.12)] 2018 9th International Conference on Computing, Communication and Networking Technologies (ICCCNT) - Image Fusion Using Convolutional Neural Network with Bilateral Filtering
摘要： Image fusion is a method of combining source images taken from the same scene. A deep convolutional neural network (CNN) is used in this paper to extract the high frequency details from the two source images. A focus map is generated after the several convolution and max-pooling layers which contains the clarity information of the source images. A fixed threshold is applied to the focus map to generate a binary segmented map which correctly classifies the pixels belonging to the focused regions. The results of binary segmentation contain some mis-classified pixels which is improved by applying a small region removal strategy to get the initial decision map. The proposed bilateral filter is a very efficient edge-preserving filter which smoothen the regions around the boundaries of the obtained decision map. The pixel-wise weighted average strategy is calculated to get the fused image with high visual quality. Experimental results show that the proposed CNN-based method produces more natural effect of the fused image.
关键词： Convolutional neural network,bilateral filter,Image fusion
更新于2025-09-11 14:15:04
-
[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 - Single Frame Super Resolution with Convolutional Neural Network for Remote Sensing Imagery
摘要： In this paper, a new convolutional neural networks based super-resolution (SR) is proposed. SR has been a hot research area for decades, and it includes two types: single frame based SR and multi-frame based SR. The focus of the paper is to reconstruct the corresponding high resolution image from a given low resolution image. The popular end-to-end learning architecture is improved and no preprocessing and image aggregation are needed. Our network model (RSCNN) uses different convolution kernels for a set of feature maps in the feature mapping step, which ensures the accuracy of reconstruction results under the premise of improving the reconstruction quality. The method is applied to Jilin-1 which is the first self-developed commercial remote sensing satellite group in China. The results show the superiority of our method both visually and numerically by comparing with other excellent image super resolution algorithms.
关键词： Jinlin-1 satellite,Single image super resolution,Convolutional Neural Network
更新于2025-09-11 14:15:04
-
[IEEE IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Valencia, Spain (2018.7.22-2018.7.27)] IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Semi-Supervised Scene Classification for Remote Sensing Images Based on CNN and Ensemble Learning
摘要： The special characteristic of remote sensing (RS) images being large scale while only low number of labeled samples available in practical applications has been obstacle to the development of RS image classification. In this paper, a novel semi-supervised framework is proposed. The high-capacity convolutional neural networks (CNN) are adopted to extract preliminary image features. The strategy of ensemble learning is then utilized to establish discriminative image representations by exploring intrinsic information of available data. Plain supervised learning is finally performed to obtain classification results. To verify the efficacy of our work, we compare it with mainstream feature representation and semi-supervised approaches. Experimental results show that by utilizing CNN features and ensemble learning, our framework can obtain more effective image representations and achieve superior results compared with other paradigms of semi-supervised classification.
关键词： convolutional neural network,ensemble learning,remote sensing images,Semi-supervised classification,scene classification
更新于2025-09-11 14:15:04
-
Generating Image Distortion Maps Using Convolutional Autoencoders with Application to No Reference Image Quality Assessment
摘要： We present two contributions in this work: (i) a reference-free image distortion map generating algorithm for spatially localizing distortions in a natural scene, and (ii) no reference image quality assessment (NRIQA) algorithms derived from the generated distortion map. We use a convolutional autoencoder (CAE) for distortion map generation. We rely on distortion maps generated by the SSIM image quality assessment (IQA) algorithm as the “ground truth” for training the CAE. We train the CAE on a synthetically generated dataset composed of pristine images and their distorted versions. Specifically, the dataset was created by applying standard distortions such as JPEG compression, JP2K compression, Additive White Gaussian Noise (AWGN) and blur to the pristine images. SSIM maps are then generated on a per distorted image basis for each of the distorted images in the dataset and are in turn used for training the CAE. We first qualitatively demonstrate the robustness of the proposed distortion map generation algorithm over several images with both traditional and authentic distortions. We also demonstrate the distortion map’s effectiveness quantitatively on both standard distortions and authentic distortions by deriving three different NRIQA algorithms. We show that these NRIQA algorithms deliver competitive performance over traditional databases like LIVE Phase II, CSIQ, TID 2013, LIVE MD and MDID 2013, and databases with authentic distortions like LIVE Wild and KonIQ-10K. In summary, the proposed method generates high quality distortion map that are used to design robust NRIQA algorithms. Further, the CAE based distortion maps generation method can easily be modified to work with other ground truth distortion maps.
关键词： Convolutional neural network,no reference image quality assessment (IQA),human visual system (HVS),autoencoders
更新于2025-09-11 14:15:04
-
Deep Visual Saliency on Stereoscopic Images
摘要： Visual saliency on stereoscopic 3D (S3D) images has been shown to be heavily influenced by image quality. Hence, this dependency is an important factor in image quality prediction, image restoration and discomfort reduction, but it is still very difficult to predict such a nonlinear relation in images. In addition, most algorithms specialized in detecting visual saliency on pristine images may unsurprisingly fail when facing distorted images. In this paper, we investigate a deep learning scheme named Deep Visual Saliency (DeepVS) to achieve a more accurate and reliable saliency predictor even in the presence of distortions. Since visual saliency is influenced by low-level features (contrast, luminance and depth information) from a psychophysical point of view, we propose seven low-level features derived from S3D image pairs and utilize them in the context of deep learning to detect visual attention adaptively to human perception. During analysis, it turns out that the low-level features play a role to extract distortion and saliency information. To construct saliency predictors, we weight and model the human visual saliency through two different network architectures, a regression and a fully convolutional neural networks (CNNs). Our results from thorough experiments confirm that the predicted saliency maps are up to 70 % correlated with human gaze patterns, which emphasize the need for the hand-crafted features as input to deep neural networks in S3D saliency detection.
关键词： deep learning,convolutional neural network,stereoscopic image,Saliency prediction,distorted image
更新于2025-09-11 14:15:04
-
[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) - Infrared Image Colorization Using a S-Shape Network
摘要： This paper proposes a novel approach for colorizing near infrared (NIR) images using a S-shape network (SNet). The proposed approach is based on the usage of an encoder-decoder architecture followed with a secondary assistant network. The encoder-decoder consists of a contracting path to capture context and a symmetric expanding path that enables precise localization. The assistant network is a shallow encoder-decoder to enhance the edge and improve the output, which can be trained end-to-end from a few image examples. The trained model does not require any user guidance or a reference image database. Furthermore, our architecture will preserve clear edges within NIR images. Our overall architecture is trained and evaluated on a real-world dataset containing a significant amount of road scene images. This dataset was captured by a NIR camera and a corresponding RGB camera to facilitate side-by-side comparison. In the experiments, we demonstrate that our SNet works well, and outperforms contemporary state-of-the-art approaches.
关键词： Colorization,Infrared,S-shape network,Convolutional neural network
更新于2025-09-10 09:29:36
-
[IEEE 2018 IEEE International Conference on Electro/Information Technology (EIT) - Rochester, MI (2018.5.3-2018.5.5)] 2018 IEEE International Conference on Electro/Information Technology (EIT) - A Survey of Traffic Sign Recognition Systems Based on Convolutional Neural Networks
摘要： In this paper, we briefly discuss the applications of Convolutional Neural Networks (CNNs) model to traffic sign recognition (TSR) systems. Traditionally, the TSRs have used different techniques to detect and classify visual data. The CNNs have been used separately to extract features and train the classifier as well as simultaneously for detection and classification tasks. One model that has been successful is the Fast Branch CNN model, which imitates biological mechanisms to become more efficient. While it is not the most accurate of the ones presented in this paper, the efficiency it exhibits under time-sensitive conditions is worth exploring because of the potential applications of such technology. The Fast Branch CNN model challenged the assumptions of past models, and this technology can only advance further if new models attempt to do the same.
关键词： CNN (Convolutional Neural Network),TSR (Traffic Sign Recognition),Classification,Detection
更新于2025-09-10 09:29:36
-
[IEEE IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Valencia, Spain (2018.7.22-2018.7.27)] IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Avalanche Detection in Sar Images Using Deep Learning
摘要： Detection of avalanches is critical for keeping avalanche inventories and management of emergency situations. In this paper we propose a deep-learning based avalanche detection method for SAR images. We utilize an existing method for proposing candidate regions, based on change detection in SAR images from multiple passes over the same area. Then a convolutional neural network is used to classify whether the candidate regions contain an avalanche or not. The proposed methodology applies existing pre-trained network that has been trained for classification of natural RGB images. SAR images represent non-standard images and we propose a method for adapting SAR images to be used in pre-trained networks for RGB images. The pre-trained network is then fine-tuned to the task of discriminating avalanches from lookalikes in the candidate regions from the SAR images. Using cross-validation, we find that the proposed method has an average classification error rate of 3.5%.
关键词： avalanche detection,convolutional neural network,deep learning,change detection,SAR images
更新于2025-09-10 09:29:36
-
[IEEE IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Valencia, Spain (2018.7.22-2018.7.27)] IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Small Sample Learning Optimization for Resnet Based Sar Target Recognition
摘要： Deep convolutional neural network (CNN) is an important branch of deep learning. Due to its strong ability of feature extraction, CNN models have been introduced to solve the problems of synthetic aperture radar automatic target recognition (SAR-ATR). However, labeled SAR images are difficult to acquire. Therefore, how to obtain a good recognition result from a small sample dataset is what we mainly focus on. In theory, a deeper network can bring a better training result. But it also brings more difficulties to the training process, especially with limited labeled training data. The residual learning which proposed in recent years can alleviate this problem effectively. In this paper, we use a deep residual network, and introduce the dropout layer into the building block to alleviate overfitting caused by limited SAR data. In order to improve the training effect, the new loss function center loss is adopted and combined with softmax loss as the supervision signal to train the deep CNN. The experimental results show that our method can achieve the classification accuracy of 99.67% with all training data, without data augmentation or pre-training. When data of the training dataset was reduced to 20%, we can still achieve a recognition result higher than 94%.
关键词： center loss,automatic target recognition (ATR),limited labeled data,Convolutional neural network (CNN),synthetic aperture radar (SAR),residual learning
更新于2025-09-10 09:29:36
-
[IEEE 2018 40th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) - Honolulu, HI, USA (2018.7.18-2018.7.21)] 2018 40th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) - OCT Fluid Segmentation using Graph Shortest Path and Convolutional Neural Network <sup>*</sup>
摘要： Diagnosis and monitoring of retina diseases related to pathologies such as accumulated fluid can be performed using optical coherence tomography (OCT). OCT acquires a series of 2D slices (Bscans). This work presents a fully-automated method based on graph shortest path algorithms and convolutional neural network (CNN) to segment and detect three types of fluid including sub-retinal fluid (SRF), intra-retinal fluid (IRF) and pigment epithelium detachment (PED) in OCT Bscans of subjects with age-related macular degeneration (AMD) and retinal vein occlusion (RVO) or diabetic retinopathy. The proposed method achieves an average dice coefficient of 76.44%, 92.25% and 82.14% in Cirrus, Spectralis and Topcon datasets, respectively. The effectiveness of the proposed methods was also demonstrated in segmenting fluid in OCT images from the 2017 Retouch challenge.
关键词： optical coherence tomography,retinal vein occlusion,fluid segmentation,graph shortest path,convolutional neural network,age-related macular degeneration
更新于2025-09-10 09:29:36