Weakly supervised target detection in remote sensing images based on transferred deep features and negative bootstrapping

Target detection in remote sensing images (RSIs) is a fundamental yet challenging problem faced for remote sensing images analysis. More recently, weakly supervised learning, in which training sets require only binary labels indicating whether an image contains the object or not, has attracted considerable attention owing to its obvious advantages such as alleviating the tedious and time consuming work of human annotation. Inspired by its impressive success in computer vision field, in this paper, we propose a novel and effective framework for weakly supervised target detection in RSIs based on transferred deep features and negative bootstrapping. On one hand, to effectively mine information from RSIs and improve the performance of target detection, we develop a transferred deep model to extract high-level features from RSIs, which can be achieved by pre-training a convolutional neural network model on a large-scale annotated dataset (e.g. ImageNet) and then transferring it to our task by domain-specifically fine-tuning it on RSI datasets. On the other hand, we integrate negative bootstrapping scheme into detector training process to make the detector converge more stably and faster by exploiting the most discriminative training samples. Comprehensive evaluations on three RSI datasets and comparisons with state-of-the-art weakly supervised target detection approaches demonstrate the effectiveness and superiority of the proposed method.     

基于损失加权修正的舰船目标HRRP小样本元学习识别方法

针对现有小样本高分辨距离像（high resolution range profile,HRRP）元学习识别方法难以适应任务经验差异的问题,提出了基于损失加权修正的舰船目标元学习识别方法。该方法以元学习理论为基础,设计了基础学习器与元学习器相结合的预训练模型。由于不同的特性损失可反映出学习经验的差异程度,故基于任务损失值对元学习器的损失函数进行加权处理,以减轻不同任务的偏差影响。然后,利用预训练模型对仿真数据的学习经验,在小样本测试任务集上进行舰船目标实测HRRP的分类识别。实验结果表明,所提方法与对比模型相比,可在小样本条件下获得更佳的识别效果,具备良好的小样本分类识别能力。     

情感识别中的迁移学习问题综述

情感识别是实现自然人机交互的必要过程。然而,情感数据高昂的采集和标注成本成为了限制情感识别研究发展的一大瓶颈。在无标注或有限标注的场景下,利用知识的跨领域或跨任务迁移提升情感识别效果的问题值得探索。本文对情感识别中的迁移学习问题进行了梳理和分析。首先,将迁移学习问题划分为针对领域差异和针对任务差异的两大部分,并进一步将每部分问题细分为多种不同的情况。随后,基于情感识别领域的研究现状,分别总结不同情况下的现有工作。在目标领域训练资源匮乏的情况下,可以利用其他带标注的数据集作为源领域训练模型,并对齐不同领域下的特征分布,或将特征映射到域间共享的空间。考虑到情感标签所提供的监督信息往往较为有限,为了进一步提升模型的识别效果,可以引入其他相关任务进行联合训练,或将预训练模型、外部知识库提供的先验语义知识迁移到情感识别任务中。最后,讨论了情感识别领域中未来需要得到更多关注和探索的迁移学习问题,旨在为研究者带来新的启发。     

Ideal AFROC and FROC Observers

Detection of multiple lesions in images is a medically important task and free-response receiver operating characteristic (FROC) analyses and its variants, such as alternative FROC (AFROC) analyses, are commonly used to quantify performance in such tasks. However, ideal observers that optimize FROC or AFROC performance metrics have not yet been formulated in the general case. If available, such ideal observers may turn out to be valuable for imaging system optimization and in the design of computer aided diagnosis techniques for lesion detection in medical images. In this paper, we derive ideal AFROC and FROC observers. They are ideal in that they maximize, amongst all decision strategies, the area, or any partial area, under the associated AFROC or FROC curve. Calculation of observer performance for these ideal observers is computationally quite complex. We can reduce this complexity by considering forms of these observers that use false positive reports derived from signal-absent images only. We also consider a Bayes risk analysis for the multiple-signal detection task with an appropriate definition of costs. A general decision strategy that minimizes Bayes risk is derived. With particular cost constraints, this general decision strategy reduces to the decision strategy associated with the ideal AFROC or FROC observer.      

Sparse representation for color image restoration

Sparse representations of signals have drawn considerable interest in recent years. The assumption that natural signals, such as images, admit a sparse decomposition over a redundant dictionary leads to efficient algorithms for handling such sources of data. In particular, the design of well adapted dictionaries for images has been a major challenge. The K-SVD has been recently proposed for this task and shown to perform very well for various grayscale image processing tasks. In this paper, we address the problem of learning dictionaries for color images and extend the K-SVD-based grayscale image denoising algorithm that appears in. This work puts forward ways for handling nonhomogeneous noise and missing information, paving the way to state-of-the-art results in applications such as color image denoising, demosaicing, and inpainting, as demonstrated in this paper.     

基于深度学习的医疗图像分割综述

自2006年深度学习这一概念提出以来,各研究领域对于深度学习技术的研究热度一直高居不下.深度学习的出现,对计算机视觉领域的发展起到了重要推动作用.计算机视觉的主要研究任务是对图像、视频等进行目标的检测、识别以及分割等,目前已经广泛应用于医疗、金融和工业领域中.其中最常见的应用场景是医学图像处理.图像分割是医学图像处理任...     

Wavelet-Domain Medical Image Denoising Using Bivariate Laplacian Mixture Model

In this paper, we proposed novel noise reduction algorithms that can be used to enhance image quality in various medical imaging modalities such as magnetic resonance and multidetector computed tomography. The noisy captured 3-D data are first transformed by discrete complex wavelet transform. Using a nonlinear function, we model the data as the sum of the clean data plus additive Gaussian or Rayleigh noise. We use a mixture of bivariate Laplacian probability density functions for the clean data in the transformed domain. The MAP and minimum mean-squared error (MMSE) estimators allow us to efficiently reduce the noise. The employed prior distribution is mixture and bivariate, and thus accurately characterizes the heavy-tail distribution of clean images and exploits the interscale properties of wavelets coefficients. In addition, we estimate the parameters of the model using local information; as a result, the proposed denoising algorithms are spatially adaptive, i.e., the intrascale dependency of wavelets is also well exploited in the enhancement process. The proposed approach results in significant noise reduction while the introduced distortions are not noticeable as a result of accurate statistical modeling. The obtained shrinkage functions have closed form, are simple in implementation, and efficiently enhances data. Our experiments on CT images show that among our derived shrinkage functions usually BiLapGausMAP produces images with higher peak SNR. However, BiLapGausMMSE is preferred especially for CT images, which have high SNRs. Furthermore, BiLapRayMAP yields better noise reduction performance for low SNR MR datasets such as high-resolution whole heart imaging while BiLapGauMAP results in better performance in MR data with higher intrinsic SNR such as functional cine data.      

基于注意力机制与迁移学习的乳腺钼靶肿块分类

针对乳腺钼靶图像中良恶性肿块难以诊断的问题,提出一种基于注意力机制与迁移学习的乳腺钼靶肿块分类方法,并用于医学影像中乳腺钼靶肿块的良恶性分类.首先,构建一种新的网络模型,该模型将注意力机制CBAM(Convolutional Block Attention Module)与残差网络ResNet50相结合,用于提高网络对...     

迁移学习用于电磁目标识别

迁移学习技术可以利用经验信息辅助当前任务,已在计算机视觉和语音识别领域得到广泛应用,但在电磁领域还没有取得明显的成就.电磁环境变化速度快,源数据或分类器模型在新环境中性能会显著下降,重新训练不仅需要大量的数据且费时费力.迁移学习技术与电磁目标识别任务十分相关,本文采用实测电磁目标数据集,探索迁移学习在解决电磁目标小样本...     

基于迁移学习和通道注意力的遥感图像场景分类

针对遥感图像场景分类任务中训练样本数量少及遥感图像背景复杂等问题,本文将迁移学习和通道注意力引入到卷积神经网络(convolutional neural network,CNN) 中,提出基于迁移学习和通道注意力的遥感图像场景分类方法。该方法首先选用经过ImageNet自然数据集预训练的两个CNN作为主干,同时引入通道注意力机制,自适应地增强主要特征,抑制次要特征;然后融合这两个网络提取的特征进行分类;最后采用微调迁移学习的方式实现目标域上的学习与分类。提出的方法在几个经典的公共数据集上进行了评估,实验结果证明了本文提出的方法在遥感图像场景分类中达到与其他先进方法相当的性能。     

A framework for medical image retrieval using machine learning and statistical similarity matching techniques with relevance feedback.

A content-based image retrieval (CBIR) framework for diverse collection of medical images of different imaging modalities, anatomic regions with different orientations and biological systems is proposed. Organization of images in such a database (DB) is well defined with predefined semantic categories; hence, it can be useful for category-specific searching. The proposed framework consists of machine learning methods for image prefiltering, similarity matching using statistical distance measures, and a relevance feedback (RF) scheme. To narrow down the semantic gap and increase the retrieval efficiency, we investigate both supervised and unsupervised learning techniques to associate low-level global image features (e.g., color, texture, and edge) in the projected PCA-based eigenspace with their high-level semantic and visual categories. Specially, we explore the use of a probabilistic multiclass support vector machine (SVM) and fuzzy c-mean (FCM) clustering for categorization and prefiltering of images to reduce the search space. A category-specific statistical similarity matching is proposed in a finer level on the prefiltered images. To incorporate a better perception subjectivity, an RF mechanism is also added to update the query parameters dynamically and adjust the proposed matching functions. Experiments are based on a ground-truth DB consisting of 5000 diverse medical images of 20 predefined categories. Analysis of results based on cross-validation (CV) accuracy and precision-recall for image categorization and retrieval is reported. It demonstrates the improvement, effectiveness, and efficiency achieved by the proposed framework.     

Towards Human-Centred Intelligent Systems — The Intelligent Assistant

In this paper, we describe some abstract features of human/machine interaction systems that are required for the production of intelligent behaviour. We introduce a subset of intelligent systems called human-centered intelligent systems (HCIS) and argue that such systems must be autonomous, robust and adaptive in order to be intelligent. We also propose soft computing as a promising new technique that can be used to build HCIS, and present examples where this is already being done. The paper defines flexibility to be a combination of the often-conflicting requirements of robustness and adaptability, and based on this we claim that the right balance between these two features is necessary to achieve intelligent behaviour.We describe the intelligent assistant (IA) system and its various components which automatically perform helpful tasks for its user, so as to enable the user to improve productivity. These tasks include time, information and communication management. Time management involves planning and scheduling, decision making and learning user habits. Information management involves information seeking and filtering, information fusion, decision making and learning user preferences. Communication management involves recognising user behaviour and learning user priorities. All these tasks depend on many factors including the type of activity, its originator, the mood of the user, past experience, and the priority of the task. The IA uses a multimodal interface with conventional interfaces such as keyboard and mouse enhanced to include vision, speech and natural language processing. The inclusion of such extra modalities extends the capabilities of existing systems at the cost of introducing extra complexity. The IA is 'smart' because it has the knowledge about tasks and the capability to learn and adapt to new interactions with its user and with other systems.     

GA-based DCT quantisation table design procedure for medical images

Medical images are widely used in the diagnosis of diseases. These imaging modalities include computerised tomography (CT), magnetic resonance imaging (MRI), ultrasonic (US) imaging, X-radiographs, etc. However, medical images have large storage requirements when high resolution is demanded; therefore, they need to be compressed to reduce the data size so as to achieve a low bit rate for transmission or storage, while maintaining image information. The Joint Photographic Experts Group (JPEG) developed an image compression tool that is one of the most widely used products for image compression. One of the factors influencing the performance of JPEG compression is the quantisation table. The bit rate and the decoded quality are determined simultaneously by the quantisation table, and therefore, the table has a strong influence on the whole compression performance. The author aims to provide a design procedure to seek sets of better quantisation parameters to raise the compression performance to achieve a lower bit rate while preserving high decoded quality. A genetic algorithm (GA) was employed to promote higher compression performance for medical images. The goal was to develop a design procedure to find quantisation tables that contribute to better compression efficiency in terms of bit rate and decoded quality. Simulations were carried out on different kinds of medical images. Resulting experimental data demonstrate that the GA-based search procedures can generate better performance than JPEG 2000 and JPEG even though the training images have different features. Additionally, if existing published quantisation tables are put into the crossover pool in the proposed GA-based system, it can improve the performance by yielding better quantisation tables.     

Investigating the Effects of Transfer Learning on ROI-based Classification of Chest CT Images: A Case Study on Diffuse Lung Diseases

Research on Computer-Aided Diagnosis (CAD) of medical images has been actively conducted to support decisions of radiologists. Since deep learning has shown distinguished abilities in classification, detection, segmentation, etc. in various problems, many studies on CAD have been using deep learning. One of the reasons behind the success of deep learning is the availability of large application-specific annotated datasets. However, it is quite tough work for radiologists to annotate hundreds or thousands of medical images for deep learning, and thus it is difficult to obtain large scale annotated datasets for various organs and diseases. Therefore, many techniques that effectively train deep neural networks have been proposed, and one of the techniques is transfer learning. This paper focuses on transfer learning and especially conducts a case study on ROI-based opacity classification of diffuse lung diseases in chest CT images. The aim of this paper is to clarify what characteristics of the datasets for pre-training and what kinds of structures of deep neural networks for fine-tuning contribute to enhance the effectiveness of transfer learning. In addition, the numbers of training data are set at various values and the effectiveness of transfer learning is evaluated. In the experiments, nine conditions of transfer learning and a method without transfer learning are compared to analyze the appropriate conditions. From the experimental results, it is clarified that the pre-training dataset with more (various) classes and the compact structure for fine-tuning show the best accuracy in this work.

     

Initialization, noise, singularities, and scale in height ridge traversal for tubular object centerline extraction

The extraction of the centerlines of tubular objects in two and three-dimensional images is a part of many clinical image analysis tasks. One common approach to tubular object centerline extraction is based on intensity ridge traversal. In this paper, we evaluate the effects of initialization, noise, and singularities on intensity ridge traversal and present multiscale heuristics and optimal-scale measures that minimize these effects. Monte Carlo experiments using simulated and clinical data are used to quantify how these "dynamic-scale" enhancements address clinical needs regarding speed, accuracy, and automation. In particular, we show that dynamic-scale ridge traversal is insensitive to its initial parameter settings, operates with little additional computational overhead, tracks centerlines with subvoxel accuracy, passes branch points, and handles significant image noise. We also illustrate the capabilities of the method for medical applications involving a variety of tubular structures in clinical data from different organs, patients, and imaging modalities.     

Estimation of the Hemodynamic Response during Motor Imagery Using Bayesian RBF Neural Network

Hemodynamic response during motor imagery(MI)b studied extensively by functional magnetic resonance imaging(fMRI)technologies.To further understand the human brain functions under MI,a more precise classification of the brain regions corresponding to each brain function b desired.In this study,a Bayesian trained radial basis function(RBF)neural network,which determines the weights and regularization parameters automatically by Bayesian learning,is applied to make a precise classification of the hemodynamic response to the tasks during the MI experiment.To illustrate the proposed method,data with MI task performance from 1 subject was used.The results demonstrate that this approach splits the hemodynamic response to different tasks successfully.     

利用无监督预训练的轨迹深度关联

针对时空相似度算法关联轨迹的局限性,采用深度学习方法进行轨迹关联,并提出了一种基于无监督预训练的匹配神经网络训练方式。利用Geohash向量嵌入对轨迹信号做特征工程处理,构建自注意力机制神经网络结构,使用无标注轨迹数据基于遮蔽预测任务进行模型预训练;然后构建孪生匹配网络结构,加载预训练模型参数;最后使用标注轨迹对数据基于均方差损失函数微调预训练模型参数得到轨迹对匹配模型。采用Geolife GPS轨迹数据集作为评估数据集进行模型训练与测试,实验结果显示,利用无监督预训练的轨迹关联方法较现有最优算法匹配准确率提高了5个百分点,达到了96.3%,充分证明了该方法的有效性。目前轨迹关联领域基于深度学习预训练模型的研究较少,该方法具有重要的参考意义。