Automated detection of proliferative diabetic retinopathy using a modified line operator and dual classification

Proliferative diabetic retinopathy (PDR) is a condition that carries a high risk of severe visual impairment. The hallmark of PDR is neovascularisation, the growth of abnormal new vessels. This paper describes an automated method for the detection of new vessels in retinal images. Two vessel segmentation approaches are applied, using the standard line operator and a novel modified line operator. The latter is designed to reduce false responses to non-vessel edges. Both generated binary vessel maps hold vital information which must be processed separately. This is achieved with a dual classification system. Local morphology features are measured from each binary vessel map to produce two separate feature sets. Independent classification is performed for each feature set using a support vector machine (SVM) classifier. The system then combines these individual classification outcomes to produce a final decision. Sensitivity and specificity results using a dataset of 60 images are 0.862 and 0.944 respectively on a per patch basis and 1.00 and 0.90 respectively on a per image basis.     

Automated segmentation of macular layers in OCT images and quantitative evaluation of performances

Optical coherence tomography (OCT) allows high-resolution and noninvasive imaging of the structure of the retina in humans. This technique revolutionized the diagnosis of retinal diseases in routine clinical practice. Nevertheless, quantitative analysis of OCT scans is yet limited to retinal thickness measurements. We propose a novel automated method for the segmentation of eight retinal layers in these images. Our approach is based on global segmentation algorithms, such as active contours and Markov random fields. Moreover, a Kalman filter is designed in order to model the approximate parallelism between the photoreceptor segments and detect them. The performance of the algorithm was tested on a set of retinal images acquired in-vivo from healthy subjects. Results have been compared with manual segmentations performed by five different experts, and intra and inter-physician variability has been evaluated as well. These comparisons have been carried out directly via the computation of the root mean squared error between the segmented interfaces, region-oriented analysis, and retrospectively on the thickness measures derived from the segmentations. This study was performed on a large database including more than seven hundred images acquired from more than one hundred healthy subjects.     

结合PCANet与线性判别分析的视网膜光学相干断层扫描图像分类

目的 主成分分析网络（PCANet）能提取图像的纹理特征,线性判别分析（LDA）提取的特征有类别区分性。本文结合这两种方法的优点,提出一种带线性判别分析的主成分分析网络（PCANet-LDA）,用于视网膜光学相干断层扫描（OCT）图像中的老年性黄斑变性（AMD）、糖尿病性黄斑水肿（DME）及正常（NOR）这3类的全自动分类。方法 PCANet-LDA算法是在PCANet的基础上添加了LDA监督层,该层加入了类标签对特征进行监督投影。首先,对OCT视网膜图像进行去噪、二值化及对齐裁剪等一系列预处理,获得感兴趣的视网膜区域;然后,将预处理图像送入一个两层的PCA卷积层,训练PCA滤波器组并提取图像的PCA特征;接着,将PCA特征送入一个非线性输出层,通过二值散列和块直方图等处理,得到图像的特征;之后,将带有类标签的图像特征送入一个LDA监督层,学习LDA矩阵并用其对图像特征进行投影,使特征具有类别区分性;最后,将投影的特征送入线性支持向量机（SVM）中对分类器进行训练和分类。结果 实验分别在医院临床数据集和杜克数据集上进行,先对OCT图像预处理进行前后对比实验,然后对PCANet特征提取的有效性进行分析,最后对PCANet算法、ScSPM算法以及提出的PCANet-LDA3种分类算法的分类效果进行对比实验。在临床数据集上,PCANet-LDA算法的总体分类正确率为97.20%,高出PCANet算法3.77%,且略优于ScSPM算法;在杜克数据集上,PCANet-LDA算法的总体分类正确率为99.52%,高出PCANet算法1.64%,略优于ScSPM算法。结论 PCANet-LDA算法的分类正确率明显高于PCANet,且优于目前用于2D视网膜OCT图像分类的先进的ScSPM算法。因此,提出的PCANet-LDA算法在视网膜OCT图像的分类上是有效且先进的,可作为视网膜OCT图像分类的基准算法。     

Automated insect identification through concatenated histograms of local appearance features: feature vector generation and region detection for deformable objects

This paper describes a computer vision approach to automated rapid-throughput taxonomic identification of stonefly larvae. The long-term objective of this research is to develop a cost-effective method for environmental monitoring based on automated identification of indicator species. Recognition of stonefly larvae is challenging because they are highly articulated, they exhibit a high degree of intraspecies variation in size and color, and some species are difficult to distinguish visually, despite prominent dorsal patterning. The stoneflies are imaged via an apparatus that manipulates the specimens into the field of view of a microscope so that images are obtained under highly repeatable conditions. The images are then classified through a process that involves (a) identification of regions of interest, (b) representation of those regions as SIFT vectors (Lowe, in Int J Comput Vis 60(2):91–110, 2004) (c) classification of the SIFT vectors into learned “features” to form a histogram of detected features, and (d) classification of the feature histogram via state-of-the-art ensemble classification algorithms. The steps (a) to (c) compose the concatenated feature histogram (CFH) method. We apply three region detectors for part (a) above, including a newly developed principal curvature-based region (PCBR) detector. This detector finds stable regions of high curvature via a watershed segmentation algorithm. We compute a separate dictionary of learned features for each region detector, and then concatenate the histograms prior to the final classification step. We evaluate this classification methodology on a task of discriminating among four stonefly taxa, two of which, Calineuria and Doroneuria, are difficult even for experts to discriminate. The results show that the combination of all three detectors gives four-class accuracy of 82% and three-class accuracy (pooling Calineuria and Doro-neuria) of 95%. Each region detector makes a valuable contribution. In particular, our new PCBR detector is able to discriminate Calineuria and Doroneuria much better than the other detectors.     

Modeling alternatives for fuzzy Markov chain-based classification of multitemporal remote sensing data

In this work, we investigate the application of modeling alternatives regarding fuzzy Markov chain-based, multitemporal, cascade classification of remote sensing data. From a theoretical viewpoint, alternative designs for the fuzzy Markov chain-based model are formally presented. From a pragmatic perspective, experimental results are discussed and analyzed, providing a deeper understanding of the virtues and odds of multitemporal remote sensing data classification based on fuzzy Markov chains. We claim that the key components of the fuzzy Markov chain-based, multitemporal classification model with respect to its alternative designs are the t-norm and s-norm operators, and the fuzzy aggregation function. The main objective of this paper is to investigate how a particular design may affect the classification performance. In addition, this paper aims at assessing the impact of the monotemporal classifiers’ accuracies on the quality of the multitemporal classification outcome, according to the selected design alternatives. In conclusion, this paper presents design guidelines for both the developer of image analysis systems and the designer of classification methods based on fuzzy Markov chains.     

基于AROP程序包的类Landsat遥感影像配准与正射纠正试验和精度分析

遥感影像的精确配准和正射纠正是进行图像融合、变化检测、图像镶嵌、定量遥感建模、多时相和多传感器影像协同应用的基础和前提。以美国国家航空和航天管理局下设LEDAPS(Landsat Ecosystem Disturbance Adaptive Processing System)课题组开发的配准与正射纠正程序包AROP(Automated Registration and Orthorectification Package)为例,详细阐述了其配准的原理与程序设计流程,并对其配准的精度进行了分析和评价。试验表明:AROP程序包算法能够找出足够的控制点,且控制点分布较为均匀,配准误差小于0.5个像元。误差特征表现为:扫描误差明显大于航向误差,误差的结果与影像漂移、DEM、坡度存在一定的相关性,高程和坡度是影响配准精度的主要因素之一。该程序包目前能够用于对我国CBERS影像的正射校正以及波段不匹配处理,但是对HJ卫星CCD影像数据配准还有待于进一步研究。     

A comparison of time series similarity measures for classification and change detection of ecosystem dynamics

Time series of remote sensing imagery or derived vegetation indices and biophysical products have been shown particularly useful to characterize land ecosystem dynamics. Various methods have been developed based on temporal trajectory analysis to characterize, classify and detect changes in ecosystem dynamics. Although time series similarity measures play an important role in these methods, a quantitative comparison of the similarity measures is lacking. The objective of this study was to provide an overview and quantitative comparison of the similarity measures in function of varying time series and ecosystem characteristics, such as amplitude, timing and noise effects. For this purpose, the performance was evaluated for the commonly used similarity measures (D), ranging from Manhattan (D_Man), Euclidean (D_E) and Mahalanobis (D_Mah) distance measures, to correlation (D_CC), Principal Component Analysis (PCA; D_PCA) and Fourier based (D_FFT,D_ξ,D_Fk) similarities. The quantitative comparison consists of a series of Monte-Carlo simulations based on subsets of global MODIS Normalized Difference Vegetation index (NDVI) and Enhanced Vegetation Index (EVI) and Leaf Area Index (LAI) data. Results of the simulations reveal four main groups of time series similarity measures with different sensitivities: (i) D_Man, D_E, D_PCA, D_Fk quantify the difference in time series values, (ii) D_Mah accounts for temporal correlation and non-stationarity of variance, (iii) D_CC measures the temporal correlation, and (iv) the Fourier based D_FFT and D_ξ show their specific sensitivity based on the selected Fourier components. The difference measures show relatively the highest sensitivity to amplitude effects, whereas the correlation based measures are highly sensitive to variations in timing and noise. The Fourier based measures, finally, depend highly on the signal to noise ratio and the balance between amplitude and phase dominance. The heterogeneity in sensitivity of each D stresses the importance of (i) understanding the time series characteristics before applying any classification of change detection approach and (ii) defining the variability one wants to identify/account for. This requires an understanding of the ecosystem dynamics and time series characteristics related to the baseline, amplitude, timing, noise and variability of the ecosystem time series. This is also illustrated in the quantitative comparison, where the different sensitivities of D for the NDVI, EVI, and LAI data relate specifically to the temporal characteristics of each data set. Additionally, the effect of noise and intra- and interclass variability is demonstrated in a case study based on land cover classification.     

基于颜色和重量特征的苹果在线分级系统设计

针对目前水果分级自动化程度较低且分级系统均为离线工作台方式的问题,设计了一种基于多信息融合的苹果采摘机器人在线自动分级系统。在提取苹果重量及颜色两个特征参数后,运用权重分析法对这两个特征参数进行数据融合,以实现双特征综合分级。为提高分级正确率,通过实验优化了权重系数。实验结果表明：所设计的分级系统可实现按重量和颜色双特征进行苹果分级,分级正确率达87%以上。     

Exudates and optic disk detection in retinal images of diabetic patients

Diabetic retinopathy is the progressive pathological alterations in the retinal microvasculature that very often causes blindness. Because of its clinical significance, it will be helpful to have regular cost‐effective eye screening for diabetic patients by developing algorithms to perform retinal image analysis, fundus image enhancement, and monitoring. The two cost‐effective algorithms are proposed for exudates detection and optic disk extraction aimed for retinal images classification and diagnosis assistance. They represent the effort made to offer a cost‐effective algorithm for optic disk identification, which will enable easier exudates extraction, exudates detection and retinal images classification aimed to assist ophthalmologists while making diagnoses. The proposed algorithms apply mathematical modeling, which enables light intensity levels emphasis, easier optic disk and exudates detection, efficient and correct classification of retinal images. The algorithm is robust to various appearance changes of retinal fundus images and shows very promising results. Fundus images are classified into those that are healthy and those affected by diabetes, based on the detected optic disk and exudates. The obtained results indicate that the proposed algorithm successfully and correctly classifies more than 98% of the observed retinal images because of the changes in the appearance of retinal fundus images typically encountered in clinical environments. Copyright © 2014 John Wiley & Sons, Ltd.     

Automated detection of fovea in fundus images based on vessel-free zone and adaptive Gaussian template

This study developed a computerised method for fovea centre detection in fundus images. In the method, the centre of the optic disc was localised first by the template matching method, the disc–fovea axis (a line connecting the optic disc centre and the fovea) was then determined by searching the vessel-free region, and finally the fovea centre was detected by matching the fovea template around the centre of the axis. Adaptive Gaussian templates were used to localise the centres of the optic disc and fovea for the images with different resolutions. The proposed method was evaluated using three publicly available databases (DIARETDB0, DIARETDB1 and MESSIDOR), which consisted of a total of 1419 fundus images with different resolutions. The proposed method obtained the fovea detection accuracies of 93.1%, 92.1% and 97.8% for the DIARETDB0, DIARETDB1 and MESSIDOR databases, respectively. The overall accuracy of the proposed method was 97.0% in this study.     

An ontology of slums for image-based classification

Information about rapidly changing slum areas may support the development of appropriate interventions by concerned authorities. Often, however, traditional data collection methods lack information on the spatial distribution of slum-dwellers. Remote sensing based methods could be used for a rapid inventory of the location and physical composition of slums. (Semi-)automatic detection of slums in image data is challenging, owing to the high variability in appearance and definitions across different contexts. This paper develops an ontological framework to conceptualize slums using input from 50 domain-experts covering 16 different countries. This generic slum ontology (GSO) comprises concepts identified at three levels that refer to the morphology of the built environment: the environs level, the settlement level and the object level. It serves as a comprehensive basis for image-based classification of slums, in particular, using object-oriented image analysis (OOA) techniques. This is demonstrated by with an example of local adaptation of GSO and OOA parameterization for a study area in Kisumu, Kenya. At the object level, building and road characteristics are major components of the ontology. At the settlement level, texture measures can be potentially used to represent the contrast between planned and unplanned settlements. At the environs level, factors which extend beyond the site itself are important indicators, e.g. hazards due to floods plains and marshy conditions. The GSO provides a comprehensive framework that includes all potentially relevant indicators that can be used for image-based slum identification. These characteristics may be different for other study areas, but show the applicability of the developed framework.     

基于深度卷积神经网络的糖尿病视网膜病变分期及病灶检测

针对糖尿病视网膜病变（DR）图像分辨率过大、病灶特征过于分散难以获取以及正负难易样本不平衡而导致DR分期精确率一直无法得到有效提高的问题,提出了改进的基于快速区域的卷积神经网络（Faster R-CNN）和子图分割相结合的DR分期方法。首先,使用子图分割解决视盘区域对于病灶识别的干扰问题;其次,在特征提取阶段使用深度残差网络以解决病灶在高分辨率眼底图像中占比小而导致的特征难以获取的问题;最后,在感兴趣区域（ROI）生成时采用在线困难样本挖掘（OHEM）方法解决正负难易样本不平衡的问题。在国际公开数据集EyePACS进行DR分期实验,所提方法在DR病分期中精确率0期达到94.83%,1期达到86.84%,2期达到94.00%,3期达到87.21%,4期达到82.96%。实验结果表明,改进后的Faster R-CNN能对DR图像高效分期并自动标注出病灶。     

Improved localization in a generalized Hough scheme for the detection of straight edges

Hough transforms are widely used for the location of straight edges in digital images, yet most common line parametrization schemes give no information on longitudinal localization. The generalized Hough transform goes some way to overcoming this problem. This paper studies how to improve the situation further. A trade-off between sensitivity and localization is found; in practical situations this results in significantly greater accuracy, but the important gain is a reduction in the number of ambiguities introduced by interactions between the transforms of unrelated straight edges.     

Comparison of different classifier algorithms for diagnosing macular and optic nerve diseases

Abstract: The aim of this research was to compare classifier algorithms including the C4.5 decision tree classifier, the least squares support vector machine (LS-SVM) and the artificial immune recognition system (AIRS) for diagnosing macular and optic nerve diseases from pattern electroretinography signals. The pattern electroretinography signals were obtained by electrophysiological testing devices from 106 subjects who were optic nerve and macular disease subjects. In order to show the test performance of the classifier algorithms, the classification accuracy, receiver operating characteristic curves, sensitivity and specificity values, confusion matrix and 10-fold cross-validation have been used. The classification results obtained are 85.9%, 100% and 81.82% for the C4.5 decision tree classifier, the LS-SVM classifier and the AIRS classifier respectively using 10-fold cross-validation. It is shown that the LS-SVM classifier is a robust and effective classifier system for the determination of macular and optic nerve diseases.     

远程DR会诊系统的设计和实现

本文讨论了建立远程糖尿病眼底视网膜病变筛查系统平台的三个关键技术点:信息传输方式,信息标准和信息安全。提出基于WebServices的系统设计模式,提出以HL7CDA标准作为医学信息传输标准,并结合XML数字签名技术来实现信息安全。解决了异地异构医疗平台之间的信息融合,实现远程糖尿病眼底视网膜病变的异地异步筛查工作。     

新的噪声污染灰度图像边缘检测统计方法*

针对传统算子进行边缘检测时易丢失边缘信息和在非边缘处增强噪声的缺陷,提出一种基于非参数变点统计分析的噪声图像边缘检测方法,该统计方法不但不需要图像数字特征的任何先验信息,而且对噪声污染的图像不作任何滤波处理.实验结果表明,提出的算法优于Sobel算子,并能抑制信噪较低的高斯噪声和密度较高的椒盐噪声对分割结果的影响,是一种有效的噪声污染灰度图像边缘检测方法.     

Fusion of color edge detection and color quantization for color image watermarking using principal axes analysis

In the past few years, many gray-level image watermarking schemes have been proposed, although the extension to the color case is rare and regularly accomplished by marking the image luminance, or by processing each color channel separately. This paper presents a new color image watermarking scheme that combines color edge detection and color quantization using principal axes analysis in three-dimensional color space. The watermark is hidden within the data by modifying a subset of carefully selected edge points to resist both geometric distortion and signal processing attacks. Experimental results show the robustness of the proposed scheme to resist common attacks.     

Object-oriented texture analysis for the unsupervised segmentation of biopsy images for cancer detection

Staining methods routinely used in pathology lead to similar color distributions in the biologically different regions of histopathological images. This causes problems in image segmentation for the quantitative analysis and detection of cancer. To overcome this problem, unlike previous methods that use pixel distributions, we propose a new homogeneity measure based on the distribution of the objects that we define to represent tissue components. Using this measure, we demonstrate a new object-oriented segmentation algorithm. Working with colon biopsy images, we show that this algorithm segments the cancerous and normal regions with 94.89 percent accuracy on the average and significantly improves the segmentation accuracy compared to its pixel-based counterpart.