高分辨卫星图像卷积神经网络分类模型

目的 卫星图像往往目标、背景复杂而且带有噪声,因此使用人工选取的特征进行卫星图像的分类就变得十分困难。提出一种新的使用卷积神经网络进行卫星图像分类的方案。使用卷积神经网络可以提取卫星图像的高层特征,进而提高卫星图像分类的识别率。方法 首先,提出一个包含六类图像的新的卫星图像数据集来解决卷积神经网络的有标签训练样本不足的问题。其次,使用了一种直接训练卷积神经网络模型和3种预训练卷积神经网络模型来进行卫星图像分类。直接训练模型直接在文章提出的数据集上进行训练,预训练模型先在ILSVRC（the ImageNet large scale visual recognition challenge）-2012数据集上进行预训练,然后在提出的卫星图像数据集上进行微调训练。完成微调的模型用于卫星图像分类。结果 提出的微调预训练卷积神经网络深层模型具有最高的分类正确率。在提出的数据集上,深层卷积神经网络模型达到了99.50%的识别率。在数据集UC Merced Land Use上,深层卷积神经网络模型达到了96.44%的识别率。结论 本文提出的数据集具有一般性和代表性,使用的深层卷积神经网络模型具有很强的特征提取能力和分类能力,且是一种端到端的分类模型,不需要堆叠其他模型或分类器。在高分辨卫星图像的分类上,本文模型和对比模型相比取得了更有说服力的结果。     

RBF神经网络在遥感影像分类中的应用研究

用RBF神经网络进行遥感影像分类，在网络结构设计上使RBF层与输出层的节点数都等于所要分类的类别数。用Kohonen聚类算法确定RBF中心的时候，用训练样本的均值作为初始中心，并在RBF宽度进行求取的时候进行了改进，以避免内存溢出。所设计的RBF神经网络分类模型具有结构简单、算法简洁的优点。实验结果表明，该方法用于遥感影像分类取得了较高的分类精度，具有实际应用价值。     

Cloud basis function neural network: A modified RBF network architecture for holistic facial expression recognition

The paper presents novel modifications to radial basis functions (RBFs) and a neural network based classifier for holistic recognition of the six universal facial expressions from static images. The new basis functions, called cloud basis functions (CBFs) use a different feature weighting, derived to emphasize features relevant to class discrimination. Further, these basis functions are designed to have multiple boundary segments, rather than a single boundary as for RBFs. These new enhancements to the basis functions along with a suitable training algorithm allow the neural network to better learn the specific properties of the problem domain. The proposed classifiers have demonstrated superior performance compared to conventional RBF neural networks as well as several other types of holistic techniques used in conjunction with RBF neural networks. The CBF neural network based classifier yielded an accuracy of 96.1%, compared to 86.6%, the best accuracy obtained from all other conventional RBF neural network based classification schemes tested using the same database.     

Editorial

This study investigates the potential of applying the radial basis function (RBF) neural network architecture for the classification of multispectral very high spatial resolution satellite images into 13 classes of various scales. For the development of the RBF classifiers, the innovative fuzzy means training algorithm is utilized, which is based on a fuzzy partition of the input space. The method requires only a short amount of time to select both the structure and the parameters of the RBF classifier. The new technique was applied to the area of Lake Kerkini, which is a wetland of great ecological value, located in northern Greece. Eleven experiments were carried out in total in order to investigate the performance of the classifier using different input parameters (spectral and textural) as well as different window sizes and neural network complexities. For comparison purposes the same satellite scene was classified using the maximum likelihood (MLH) classification with the same set of training samples. Overall, the neural network classifiers outperformed the MLH classification by 10–17%, reaching a maximum overall accuracy of 78%. Analysis showed that the selection of input parameters is vital for the success of the classifiers. On the other hand, the incorporation of textural analysis and/or modification of the window size do not affect the performance substantially.     

高光谱图像小样本分类的卷积神经网络方法

目的 与传统分类方法相比,基于深度学习的高光谱图像分类方法能够提取出高光谱图像更深层次的特征。针对现有深度学习的分类方法网络结构简单、特征提取不够充分的问题,提出一种堆叠像元空间变换信息的数据扩充方法,用于解决训练样本不足的问题,并提出一种基于不同尺度的双通道3维卷积神经网络的高光谱图像分类模型,来提取高光谱图像的本质空谱特征。方法 通过对高光谱图像的每一像元及其邻域像元进行旋转、行列变换等操作,丰富中心像元的潜在空间信息,达到数据集扩充的作用。将扩充之后的像素块输入到不同尺度的双通道3维卷积神经网络学习训练集的深层特征,实现更高精度的分类。结果 5次重复实验后取平均的结果表明,在随机选取了10%训练样本并通过8倍数据扩充的情况下,Indian Pines数据集实现了98.34%的总体分类精度,Pavia University数据集总体分类精度达到99.63%,同时对比了不同算法的运行时间,在保证分类精度的前提下,本文算法的运行时间短于对比算法,保证了分类模型的稳定性、高效性。结论 本文提出的基于双通道卷积神经网络的高光谱图像分类模型,既解决了训练样本不足的问题,又综合了高光谱图像的光谱特征和空间特征,提高了高光谱图像的分类精度。     

基于改进QPSO和RBF神经网络的文本分类方法

为提高文本分类的准确性,本文提出了一种基于量子PSO和RBF神经网络的新的文本分类方法.首先建立描述样本类别的关键词集合,并采用模糊向量空间模型建立每类样本的特征向量,然后采用RBF神经网络实施文本自动分类,采用改进的量子PSO优化RBF神经网络的参数,以提高其逼近能力.选取中国期刊网的部分文献作为实验数据,实验结果说明本文所提出方法的分类精准度与其他同类方法相比有明显的提高.     

基于连通区域的传真图像版面分割与分类算法

针对传真图像,为了提高版面分割与分类准确率,提高处理速度,以连通区域为处理元素,通过合理设定阈值,将水平和垂直相邻连通区域进行合并,快速准确地分割图像。并且将版面分割过程与分类过程相结合,根据连通区域的大小建立矩阵,提取能够表征区域信息的八维特征,然后使用BP神经网络将版面区域分为文字区域和非文字区域两类。实验中得到版面分割准确率为89.2%,版面分类准确率为94.22%。实验结果证明,该算法能够快速准确地对传真图像进行分割和分类,具有较强的实用价值。     

An optimized RBF neural network algorithm based on partial least squares and genetic algorithm for classification of small sample

Radial basis function (RBF) neural network can use linear learning algorithm to complete the work formerly handled by nonlinear learning algorithm, and maintain the high precision of the nonlinear algorithm. However, the results of RBF would be slightly unsatisfactory when dealing with small sample which has higher feature dimension and fewer numbers. Higher feature dimension will influence the design of neural network, and fewer numbers of samples will cause network training incomplete or over-fitted, both of which restrict the recognition precision of the neural network. RBF neural network has some drawbacks, for example, it is hard to determine the numbers, center and width of the hidden layer’s neurons, which constrain the success of training. To solve the above problems, partial least squares (PLS) and genetic algorithm(GA)are introduced into RBF neural network, and better recognition precision will be obtained, because PLS is good at dealing with the small sample data, it can reduce feature dimension and make low-dimensional data more interpretative. In addition, GA can optimize the network architecture, the weights between hidden layer and output layer of the RBF neural network can ease non-complete network training, the way of hybrid coding and simultaneous evolving is adopted, and then an accurate algorithm is established. By these two consecutive optimizations, the RBF neural network classification algorithm based on PLS and GA (PLS-GA-RBF) is proposed, in order to solve some recognition problems caused by small sample. Four experiments and comparisons with other four algorithms are carried out to verify the superiority of the proposed algorithm, and the results indicate a good picture of the PLS-GA-RBF algorithm, the operating efficiency and recognition accuracy are improved substantially. The new small sample classification algorithm is worthy of further promotion.     

Intelligent sensor system for discrimination of material type using neural networks

Neural network based classification of material type even with the variation in the sensor parameter is investigated in this paper. The sensor is developed by means of a lightweight plunger probe and an optical mouse sensor. An experimental prototype was developed which involves bouncing or hopping of the plunger based impact probe freely on the plain surface of an object under test. The experiment is conducted to obtain the bouncing signals for plain surface of an objects kept at different distances from the probe. During the bouncing of the probe, time varying signals are generated from optical mouse that are recorded in data files on PC. Some dominant unique features are then extracted using signal processing tools to optimize neural network based classifier. The time and features of bouncing signal are related to the material type, and each material has a unique set of such properties. It is found that the sensor system is intelligent due to its ability to classify the material type even with the variation in the sensor parameter (distance between the sensor probe and plain objects). The classifiers are developed using two neural networks configurations, namely a well-known Multi-layer Perceptron Neural Networks (MLP NN), and Radial Basis Function Neural Networks (RBF NN). MLP NN and RBF NN models are designed to maximize accuracy under the constraints of minimum network dimension.The optimal parameters of MLP NN and RBF NN models based on various performance measures that include percentage classification accuracy (PCLA) on the testing data, and area under Receiver Operating Characteristics (ROC), and are determined. For the sensor data set, the PCLA of both the classifiers are found reasonable consistently in respect of rigorous testing using different data partitions. The areas under the ROC curves are close to unity. Performances of the two classifiers have been compared. It has been found that the RBF NN is more robust to noise, and epochs required for training are very less as compared to that for MLP NN.     

一种高精度微孔板浑浊度识别网络

提出一种基于卷积神经网络的高精度微孔板浑浊度分类算法。该算法主要将传统图像处理技术与卷积神经网络技术相结合,通过传统图像处理算法将圆孔从自然拍摄的微孔板图像中切割下来,并将切割下来的圆孔图像制作成圆孔数据集,用于网络模型的训练、评估和测试。同时,通过深度学习技术,设计并训练多个基于深度可分离卷积核的卷积神经网络模型,然后筛选出评估准确率最高的浑浊度分类模型,应用于圆孔识别系统,从而可提高研究人员的工作效率。     

The influence of the radial basis function training strategy on the performance of a neural gripper

In this article, a neural network-based grasping system that is able to collect objects of arbitrary shape is introduced. The grasping process is split into three functional blocks: image acquisition and processing, contact point estimation, and contact force determination. The paper focuses on the second block, which contains two neural networks. A competitive Hopfield neural network first determines an approximate polygon for an object outline. These polygon edges are the input for a supervised neural network model [radial basis function (RBF) or multilayer perceptions], which then defines the contact points. Tests were conducted with objects of different shapes, and experimental results suggest that the performance of the neural gripper and its learning rate are significantly influenced by the choice of supervised training model and RBF learning algorithm.     

Combined projection and kernel basis functions for classification in evolutionary neural networks

This paper proposes a hybrid neural network model using a possible combination of different transfer projection functions (sigmoidal unit, SU, product unit, PU) and kernel functions (radial basis function, RBF) in the hidden layer of a feed-forward neural network. An evolutionary algorithm is adapted to this model and applied for learning the architecture, weights and node typology. Three different combined basis function models are proposed with all the different pairs that can be obtained with SU, PU and RBF nodes: product–sigmoidal unit (PSU) neural networks, product–radial basis function (PRBF) neural networks, and sigmoidal–radial basis function (SRBF) neural networks; and these are compared to the corresponding pure models: product unit neural network (PUNN), multilayer perceptron (MLP) and the RBF neural network. The proposals are tested using ten benchmark classification problems from well known machine learning problems. Combined functions using projection and kernel functions are found to be better than pure basis functions for the task of classification in several datasets.     

Analysis of classification accuracy for pre-filtered multichannel remote sensing data

Noise is one of the main factors degrading the quality of original multichannel remote sensing data and its presence influences classification efficiency, object detection, etc. Thus, pre-filtering is often used to remove noise and improve the solving of final tasks of multichannel remote sensing. Recent studies indicate that a classical model of additive noise is not adequate enough for images formed by modern multichannel sensors operating in visible and infrared bands. However, this fact is often ignored by researchers designing noise removal methods and algorithms. Because of this, we focus on the classification of multichannel remote sensing images in the case of signal-dependent noise present in component images. Three approaches to filtering of multichannel images for the considered noise model are analysed, all based on discrete cosine transform in blocks. The study is carried out not only in terms of conventional efficiency metrics used in filtering (MSE) but also in terms of multichannel data classification accuracy (probability of correct classification, confusion matrix). The proposed classification system combines the pre-processing stage where a DCT-based filter processes the blocks of the multichannel remote sensing image and the classification stage. Two modern classifiers are employed, radial basis function neural network and support vector machines. Simulations are carried out for three-channel image of Landsat TM sensor. Different cases of learning are considered: using noise-free samples of the test multichannel image, the noisy multichannel image and the pre-filtered one. It is shown that the use of the pre-filtered image for training produces better classification in comparison to the case of learning for the noisy image. It is demonstrated that the best results for both groups of quantitative criteria are provided if a proposed 3D discrete cosine transform filter equipped by variance stabilizing transform is applied. The classification results obtained for data pre-filtered in different ways are in agreement for both considered classifiers. Comparison of classifier performance is carried out as well. The radial basis neural network classifier is less sensitive to noise in original images, but after pre-filtering the performance of both classifiers is approximately the same.     

IIRNet: A lightweight deep neural network using intensely inverted residuals for image recognition

Deep neural networks have achieved great success in many tasks of pattern recognition. However, large model size and high cost in computation limit their applications in resource-limited systems. In this paper, our focus is to design a lightweight and efficient convolutional neural network architecture by directly training the compact network for image recognition. To achieve a good balance among classification accuracy, model size, and computation complexity, we propose a lightweight convolutional neural network architecture named IIRNet for resource-limited systems. The new architecture is built based on Intensely Inverted Residual block (IIR block) to decrease the redundancy of the convolutional blocks. By utilizing two new operations, intensely inverted residual and multi-scale low-redundancy convolutions, IIR block greatly reduces its model size and computational costs while matches the classification accuracy of the state-of-the-art networks. Experiments on CIFAR-10, CIFAR-100, and ImageNet datasets demonstrate the superior performance of IIRNet on the trade-offs among classification accuracy, computation complexity, and model size, compared to the mainstream compact network architectures.     

Probabilistic neural network for breast cancer classification

Among cancers, breast cancer causes second most number of deaths in women. To reduce the high number of unnecessary breast biopsies, several computer-aided diagnosis systems have been proposed in the last years. These systems help physicians in their decision to perform a breast biopsy on a suspicious lesion seen in a mammogram or to perform a short-term follow-up examination instead. In clinical diagnosis, the use of artificial intelligent techniques as neural networks has shown great potential in this field. In this paper, three classification algorithms, multi-layer perceptron (MLP), radial basis function (RBF) and probabilistic neural networks (PNN), are applied for the purpose of detection and classification of breast cancer. Decision making is performed in two stages: training the classifiers with features from Wisconsin Breast Cancer database and then testing. The performance of the proposed structure is evaluated in terms of sensitivity, specificity, accuracy and ROC. The results revealed that PNN was the best classifiers by achieving accuracy rates of 100 and 97.66 % in both training and testing phases, respectively. MLP was ranked as the second classifier and was capable of achieving 97.80 and 96.34 % classification accuracy for training and validation phases, respectively, using scaled conjugate gradient learning algorithm. However, RBF performed better than MLP in the training phase, and it has achieved the lowest accuracy in the validation phase.     

基于改进胶囊神经网络的遥感影像分类

针对遥感影像卷积神经网络（CNN）分类会导致特征信息丢失及泛化能力差的问题,提出一种基于通道注意力和混合注意力改进的胶囊神经网络分类模型。首先,为了胶囊神经网络能够适应于大尺寸输入图像,在特征提取模块中使用2个最大池化层;其次,为了提高分类精度,分别将SENet注意力和CBAM注意力加在特征提取模块的最后一层去改进特征提取模块;最后,将样本集随机地划分为训练集、验证集和测试集,进一步使用训练集和验证集训练模型,测试集测试模型,使用AID数据集对模型分类的泛化能力进行验证。实验结果表明:基于SENet网络改进的胶囊神经网络的准确率与Kappa系数要高于其他模型,泛化能力也优于其他模型,本文提出的模型的总体分类精度和泛化能力有了显著性提升,从而验证了本文方法的可行性和使用性。     

Lung cancer classification using neural networks for CT images

Early detection of cancer is the most promising way to enhance a patient's chance for survival. This paper presents a computer aided classification method in computed tomography (CT) images of lungs developed using artificial neural network. The entire lung is segmented from the CT images and the parameters are calculated from the segmented image. The statistical parameters like mean, standard deviation, skewness, kurtosis, fifth central moment and sixth central moment are used for classification. The classification process is done by feed forward and feed forward back propagation neural networks. Compared to feed forward networks the feed forward back propagation network gives better classification. The parameter skewness gives the maximum classification accuracy. Among the already available thirteen training functions of back propagation neural network, the Traingdx function gives the maximum classification accuracy of 91.1%. Two new training functions are proposed in this paper. The results show that the proposed training function 1 gives an accuracy of 93.3%, specificity of 100% and sensitivity of 91.4% and a mean square error of 0.998. The proposed training function 2 gives a classification accuracy of 93.3% and minimum mean square error of 0.0942.     

Document Analysis by Crosscount Approach

In this paper a new feature called crosscount for document analysis is introduced.The reature crosscount is a function of white line segment with its start on the edge of document images.It reflects not only the contour of image,but also the periodicity of white lines(background)and text lines in the document images.In complex printed-page layouts,there are different blocks such as textual,graphical,tabular,and so on.Of these blocks,textual ones have the most obvious periodicity with their homogeneous white lines arranged regularly.The important property of textual blocks can be extracted by crosscount functions.here the document layouts are classified into three classes on the basis of their physical structures.Then the definition and properties of the crosscount function are described.According to the classification of document layouts,the application of this new feature to different types of document images‘ analysis and understanding is discussed.     

轻量化卷积神经网络在SAR图像语义分割中的应用

针对合成孔径雷达图像的语义分割问题,构建了一个全新的TerraSAR-X语义分割数据集GDUT-Nansha。然后,为解决传统深度学习方法模型体积大,难以在样本数量偏少的合成孔径雷达图像数据集上应用的问题,对轻量化卷积神经网络ENet模型进行了分析和改造。提出了一种改进的轻量化卷积神经网络模型(revised weighted loss eNet,RWL-ENet);针对合成孔径雷达图像数据集样本不平衡问题,使用了带有权重的损失函数。通过和其他经典卷积神经网络语义分割模型的对比实验,验证了新数据集的可靠性;同时,在参数量和模型体积远远小于其他网络模型的前提下,RWL-ENet模型在像素精度、平均像素精度、平均交并比三个定量指标上分别达到了0.884、0.804和0.645。     

基于 DenseNet-Attention 模型的高光谱图像分类

摘 要：针对高光谱图像标记样本量少，提取特征不充分以及提取到的特征不区分贡献度 的问题，提出一个新型的 DenseNet-Attention 网络模型(DANet)。首先，该模型利用三维卷积核 同步提取联合光谱空间特征，同时密集连接网络(DenseNet)的稠密连接块除了能够充分提取更 加鲁棒的特征外，还减少了大量参数；其次，自注意力(self-attention)机制作为一个模块加入到 稠密连接块中，可以使上层提取到的特征在进入下一层网络之前，经过该模块对其进行权重分 配，使具有丰富的物类别信息的特征得到加强，进而区分特征的贡献度。网络模型以原始高光 谱图像邻域块作为输入，无需任何预处理，是一个端对端学习的深度神经网络。在印第安松树 林和帕维亚大学数据集上进行对比试验，网络模型的分类精度分别能够达到 99.43%和 99.99%， 有效提高了高光谱图像分类精度。