Effects of different preprocessing algorithms on the prognostic value of breast tumour microscopic images

The purpose of this study was to improve the prognostic value of tumour histopathology image analysis methodology by image preprocessing. Key image qualities were modified including contrast, sharpness and brightness. The texture information was subsequently extracted from images of haematoxylin/eosin‐stained tumour tissue sections by GLCM, monofractal and multifractal algorithms without any analytical limitation to predefined structures. Images were derived from patient groups with invasive breast carcinoma (BC, 93 patients) and inflammatory breast carcinoma (IBC, 51 patients). The prognostic performance was indeed significantly enhanced by preprocessing with the average AUCs of individual texture features improving from 0.68 ± 0.05 for original to 0.78 ± 0.01 for preprocessed images in the BC group and 0.75 ± 0.01 to 0.80 ± 0.02 in the IBC group. Image preprocessing also improved the prognostic independence of texture features as indicated by multivariate analysis. Surprisingly, the tonal histogram compression by the nonnormalisation preprocessing has prognostically outperformed the tested contrast normalisation algorithms. Generally, features without prognostic value showed higher susceptibility to prognostic enhancement by preprocessing whereas IDM texture feature was exceptionally susceptible. The obtained results are suggestive of the existence of distinct texture prognostic clues in the two examined types of breast cancer. The obtained enhancement of prognostic performance is essential for the anticipated clinical use of this method as a simple and cost‐effective prognosticator of cancer outcome.     

Automated system for characterization and classification of malaria‐infected stages using light microscopic images of thin blood smears

In this paper, we propose a comprehensive image characterization cum classification framework for malaria‐infected stage detection using microscopic images of thin blood smears. The methodology mainly includes microscopic imaging of Leishman stained blood slides, noise reduction and illumination correction, erythrocyte segmentation, feature selection followed by machine classification. Amongst three‐image segmentation algorithms (namely, rule‐based, Chan–Vese‐based and marker‐controlled watershed methods), marker‐controlled watershed technique provides better boundary detection of erythrocytes specially in overlapping situations. Microscopic features at intensity, texture and morphology levels are extracted to discriminate infected and noninfected erythrocytes. In order to achieve subgroup of potential features, feature selection techniques, namely, F‐statistic and information gain criteria are considered here for ranking. Finally, five different classifiers, namely, Naive Bayes, multilayer perceptron neural network, logistic regression, classification and regression tree (CART), RBF neural network have been trained and tested by 888 erythrocytes (infected and noninfected) for each features’ subset. Performance evaluation of the proposed methodology shows that multilayer perceptron network provides higher accuracy for malaria‐infected erythrocytes recognition and infected stage classification. Results show that top 90 features ranked by F‐statistic (specificity: 98.64%, sensitivity: 100%, PPV: 99.73% and overall accuracy: 96.84%) and top 60 features ranked by information gain provides better results (specificity: 97.29%, sensitivity: 100%, PPV: 99.46% and overall accuracy: 96.73%) for malaria‐infected stage classification.     

Combustion monitoring of a water tube boiler using a discriminant radial basis network

This research work includes a combination of Fisher’s linear discriminant (FLD) analysis and a radial basis network (RBN) for monitoring the combustion conditions for a coal fired boiler so as to allow control of the air/fuel ratio. For this, two-dimensional flame images are required, which were captured with a CCD camera; the features of the images—average intensity, area, brightness and orientation etc of the flame—are extracted after preprocessing the images. The FLD is applied to reduce the n-dimensional feature size to a two-dimensional feature size for faster learning of the RBN. Also, three classes of images corresponding to different burning conditions of the flames have been extracted from continuous video processing. In this, the corresponding temperatures, and the carbon monoxide (CO) emissions and those of other flue gases have been obtained through measurement. Further, the training and testing of Fisher’s linear discriminant radial basis network (FLDRBN), with the data collected, have been carried out and the performance of the algorithms is presented.     

数码相机的彩色图像预处理过程

结合数码相机图像采集特性和人眼视觉误差,讨论了彩色图像预处理过程.通过分析彩色图像的数据格式,将八叉树量化方法用于彩色图像的颜色信息量化,并采用矢量中值滤波消除彩色图像噪声,以实现对彩色图像的合理预处理.结合实验讨论分析了矢量中值滤波法的滤波器窗口参数,并结合颜色信息相似度理论提出了彩色图像预处理效果的评价方法.实验获得了合理的颜色量化方法和滤波器窗口参数,处理得到的图像最大色差变化控制在3个CIELAB色差单位.结果表明,用八叉树结构量化和矢量中值滤波对数码相机采集到的彩色图像进行预处理可达到较理想的效果.     

Plasmodium species aware based quantification of malaria parasitemia in light microscopy thin blood smear

Malaria is a serious worldwide disease, caused by a bite of a female Anopheles mosquito. The parasite transferred into complex life round in which it is grown and reproduces into the human body. The detection and recognition of Plasmodium species are possible and efficient through a process called staining (Giemsa). The staining process slightly colorizes the red blood cells (RBCs) but highlights Plasmodium parasites, white blood cells and artifacts. Giemsa stains nuclei, chromatin in blue tone and RBCs in pink color. It has been reported in numerous studies that manual microscopy is not a trustworthy screening technique when performed by nonexperts. Malaria parasites host in RBCs when it enters the bloodstream. This paper presents segmentation of Plasmodium parasite from the thin blood smear points on region growing and dynamic convolution based filtering algorithm. After segmentation, malaria parasite classified into four Plasmodium species: Plasmodium falciparum, Plasmodium ovale, Plasmodium vivax, and Plasmodium malaria. The random forest and K‐nearest neighbor are used for classification base on local binary pattern and hue saturation value features. The sensitivity for malaria parasitemia (MP) is 96.75% on training and testing of the proposed approach while specificity is 94.59%. Beside these, the comparisons of the two features are added to the proposed work for classification having sensitivity is 83.60% while having specificity is 94.90% through random forest classifier based on local binary pattern feature.     

基于铁谱图像异类特征融合的磨损类型识别方法

针对铁谱磨粒图像识别中存在特征单一、异类特征的综合利用率低等问题,提出一种磨粒图像多特征的异类信息融合识别方法。首先,对在线铁谱图像预处理基础上提取磨粒纹理(ASM、熵、相关、对比度)、颜色(均值、方差、斜度)、几何(7个不变矩)3种统计特征;其次,对提取特征数据进行[0,1]归一化处理,采用超球心间距法确定核参数,运用超球多类SVM实现基于单种特征的多类磨损识别;最后,在单种特征识别基础上通过后验概率构造3种特征所需的软判决基本概率赋值(BPA)函数,运用超球多类SVM与D-S证据理论结合法实现异类特征融合的铁谱图像识别。特征融合方法识别最高识别率达到了96.1%,与单一特征识别结果相比,识别准确度更高,且实现了不同特征的互补。     

Color‐based tumor tissue segmentation for the automated estimation of oral cancer parameters

This article presents an automatic color‐based feature extraction system for parameter estimation of oral cancer from optical microscopic images. The system first reduces image‐to‐image variations by means of color normalization. We then construct a database which consists of typical cancer images. The color parameters extracted from this database are then used in automated online sampling from oral cancer images. Principal component analysis is subsequently used to divide the color features into four tissue types. Each pixel in the cancer image is then classified into the corresponding tissue types based on the Mahalanobis distance. The aforementioned procedures are all fully automated; in particular, the automated sampling step greatly reduces the need for intensive labor in manual sampling and training. Experiments reveal high levels of consistency among the results achieved using the manual, semiautomatic, and fully automatic methods. Parameter comparisons between the four cancer stages are conducted, and only the mean parameters between early and late cancer stages are statistically different. In summary, the proposed system provides a useful and convenient tool for automatic segmentation and evaluation for stained biopsy samples of oral cancer. This tool can also be modified and applied to other tissue images with similar staining conditions. Microsc. Res. Tech. 2009. © 2009 Wiley‐Liss, Inc.     

An image processing pipeline to detect and segment nuclei in muscle fiber microscopic images

Muscle fiber images play an important role in the medical diagnosis and treatment of many muscular diseases. The number of nuclei in skeletal muscle fiber images is a key bio‐marker of the diagnosis of muscular dystrophy. In nuclei segmentation one primary challenge is to correctly separate the clustered nuclei. In this article, we developed an image processing pipeline to automatically detect, segment, and analyze nuclei in microscopic image of muscle fibers. The pipeline consists of image pre‐processing, identification of isolated nuclei, identification and segmentation of clustered nuclei, and quantitative analysis. Nuclei are initially extracted from background by using local Otsu's threshold. Based on analysis of morphological features of the isolated nuclei, including their areas, compactness, and major axis lengths, a Bayesian network is trained and applied to identify isolated nuclei from clustered nuclei and artifacts in all the images. Then a two‐step refined watershed algorithm is applied to segment clustered nuclei. After segmentation, the nuclei can be quantified for statistical analysis. Comparing the segmented results with those of manual analysis and an existing technique, we find that our proposed image processing pipeline achieves good performance with high accuracy and precision. The presented image processing pipeline can therefore help biologists increase their throughput and objectivity in analyzing large numbers of nuclei in muscle fiber images. Microsc. Res. Tech. 77:547–559, 2014. © 2014 Wiley Periodicals, Inc.     

颜色空间统计联合纹理特征的无参考图像质量评价

为了客观评价图像质量,本文提出联合颜色空间统计特征和权重局部二值模式(Local Binary Pattern,LBP)纹理特征的无参考图像质量评价模型。首先,对失真图像进行亮度去均值对比度归一化(Mean Subtracted Contrast Normalized,MSCN)操作得到MSCN系数;然后,对MSCN系数提取其统计参数特征和权重LBP直方图特征,其中统计参数由广义高斯模型获得,权重为MSCN系数的幅度。另外,还采用了Lαβ颜色空间下红绿和蓝黄分量的自然场景统计(Natural Scence Statistics,NSS)特征来增强基于颜色失真的描述,并运用非对称广义高斯模型获得统计参数特征。最后,运用SVR建立图像质量评价指标到主观质量得分的回归模型。在LIVE,CSIQ,TID2013和MLIVE数据库上的实验结果表明:4个数据库加权平均Spearman秩相关系数为0.776,Pearson线性相关系数为0.821,均优于其他方法;图像大小为512×512时特征提取只需0.19s。本文提出的方法与人眼主观感知具有良好的一致性,并具有复杂度低等优点。     

交通路标图像的分割与识别方法研究

对交通道路标志进行实时、正确的识别,是车辆自动导航中的一个重要方面.提出一种基于尺度不变特征变换(SIFT)算法的交通路标识别方法.该方法首先根据交通路标的色彩信息,利用HSV颜色空间对输入含有路标的彩色图像直接进行处理,从而快速分割提取目标区域;然后基于SIFT特征将分割提取的目标图像与数据库中原有图像进行特征匹配,...     

Prediction of Excess Air Factor in Automatic Feed Coal Burners by Processing of Flame Images

In this study, the relationship between the visual information gathered from the flame images and the excess air factor k in coal burners is investigated. In conventional coal burners the excess air factor k. can be obtained using very expensive air measurement instruments. The proposed method to predict k for a specific time in the coal burners consists of three distinct and consecutive stages; a) online flame images acquisition using a CCD camera, b) extraction meaningful information(flame intensity and brightness)from flame images, and c) learning these information(image features) with ANNs and estimate k. Six different feature extraction methods have been used: CDF of Blue Channel, Co-Occurrence Matrix, L_∞-Frobenius Norms,Radiant Energy Signal(RES), PCA and Wavelet. When compared prediction results, it has seen that the use of cooccurrence matrix with ANNs has the best performance(RMSE = 0.07) in terms of accuracy. The results show that the proposed predicting system using flame images can be preferred instead of using expensive devices to measure excess air factor in during combustion.     

Microscopic skin laceration segmentation and classification: A framework of statistical normal distribution and optimal feature selection

Among precision medical techniques, medical image processing is rapidly growing as a successful tool for cancer detection. Skin cancer is one of the crucial cancer types. It is identified through computer vision (CV) techniques using dermoscopic images. The early diagnosis skin cancer from dermoscopic images can be decrease the mortality rate. We propose an automated system for skin lesion detection and classification based on statistical normal distribution and optimal feature selection. Local contrast is controlled using a brighter channel enhancement technique, and segmentation is performed through a statistical normal distribution approach. The multiplication law of probability is implemented for the fusion of segmented images. In the feature extraction phase, optimized histogram, optimized color, and gray level co‐occurrences matrices features are extracted and covariance‐based fusion is performed. Subsequently, optimal features are selected through a binary grasshopper optimization algorithm. The selected optimal features are finally fed to a classifier and evaluated on the ISBI 2016 and ISBI 2017 data sets. Classification accuracy is computed using different Support Vector Machine (SVM) kernel functions, and the best accuracy is obtained for the cubic function. The average accuracies of the proposed segmentation on the PH2 and ISBI 2016 data sets are 93.79 and 96.04%, respectively, for an image size 512 × 512. The accuracies of the proposed classification on the ISBI 2016 and ISBI 2017 data sets are 93.80 and 93.70%, respectively. The proposed system outperforms existing methods on selected data sets.     

基于图像特征提取的地铁客流量自动统计系统

由于地铁客流具有动态流动性，采集的视频图像间的灰度值易发生突变，造成传统识别算法的统计误差较大。通过在传统特征提取方法的基础上，增加小波变换预处理算法，对视频图像进行多层分解及图像灰度阈值预处理，剔除灰度突变的干扰像素点，以提升特征提取精度。实验结果表明，改进型客流识别与统计算法可以很好地滤除部分干扰像素点，统计误差仅为传统方法的39.8%。     

oRGB颜色空间的夜视图像彩色融合

针对红外与可见光双波段夜视图像增强提出了基于oRGB颜色空间的彩色融合算法。选择亮度与颜色分离的对立色颜色空间,应用多尺度Top-hat变换方法将红外与可见光增强融合图像作为伪彩色图像的亮度分量,根据双波段图像的差异性并结合参考图像色调构造颜色分量;为避免线性变换的颜色传递方法容易造成灰度级压缩或过饱和情况,在分析颜色传递机理的基础上,使用非线性的直方图匹配方法实现颜色传递。实验结果表明,融合图像的颜色分布更自然,纹理细节、整体亮度和对比度等视觉特征表现更优。     

Measure of microscopy mineral image clarity constructed using the sequentially combined toggle and top‐hat based contrast operator

Correctly and effectively quantifying the clarity of microscopy mineral image is important and useful to produce clear image for mineral analysis. To well quantify the clarity of mineral image, an effective measure constructed using the sequentially combined toggle and top‐hat based contrast operator is proposed in this article. First, the sequentially combined toggle and top‐hat based contrast operator is used to extract the effective features which represent the clarity of mineral image. Second, the multiscale clarity features corresponding to each scale are extracted. Finally, after combining the multiscale clarity features in one feature image, the mean value of the feature image is constructed as the measure of mineral image clarity. Because the sequentially combined toggle and top‐hat based contrast operator could effectively extract the clarity features, the clarity features are directly used as the measure of image clarity. Therefore, the proposed measure is effective and reasonable. The experimental results on different types of microscopy mineral images verified the effective performance of the proposed measure. Microsc. Res. Tech. 76:811–817, 2013. © 2013 Wiley Periodicals, Inc.     

基于多模态信息融合的图像情感标注方法

随着互联网中图像资源的不断增长,情感作为图像的一个重要语义,是人们检索和选择图像的重要依据,因此对于图像进行情感标注显得至关重要。结合脑电信号(EEG)和图像内容,提出了一种基于多模态信息融合的图像情感标注方法。首先,提取EEG频域特征及图像特征(颜色及纹理);其次,结合两者特征信息,基于两种融合策略(特征层和决策层),构建支持向量机分类模型,进行图像情感识别与标注。为了评估方法的有效性,使用国际情绪图片系统公共数据集进行了实验验证。结果表明,提出的多模态信息融合图像情感标注方法优于单独使用EEG或图像内容的标注方法。此外,该成果有助于缩小低层视觉特征和高层情感语义之间的语义鸿沟。     

细节保留的多曝光图像融合

针对传统的多曝光图像融合算法存在的细节丟失严重和鬼影现象,提出了一种细节保留的多曝光图像融合算法。该算法首先计算曝光序列的3个特征指标:图像细节、曝光亮度和色彩信息,其中图像细节通过引导滤波计算,曝光亮度的权值由高斯方程分配,而曝光序列的色彩信息用色彩饱和度表示。然后,利用差分图和邻域相关系数检测多曝光序列中运动物体,利用3个特征指标和运动目标检测结果分别计算静态场景和动态场景的融合权值图。为了消除噪声的影响,采用递归滤波器来修正融合权值图。最后,采取加权融合的方式得到融合图像。选取10组不同的曝光序列,分别从主观和客观两方面与6种传统的融合算法进行了比较。实验结果表明,本文算法保留了丰富的细节信息,呈现出了更加生动自然的现实场景,并且有效去除了由运动物体产生的鬼影现象,效果优于其他比较算法,在静态场景和动态场景的曝光融合中都取得了好的效果。     

模拟生物视觉机制的彩色人脸识别方法

为了充分利用彩色图像提供的信息提高人脸识别的性能,提出了一种模拟生物视觉机制的彩色人脸识别方法。该方法首先构造一种模拟人类的色彩感知机制的对立色模型,将彩色人脸图像描述为对立色形式。然后,模拟初级视皮层的信息处理机制,从图像对立色描述的亮度分量和色度分量分别提取人脸的纹理特征和色彩特征。最后,分别对纹理特征和色彩特征进行分类识别,并将二者的识别相似度融合得到最终的人脸识别结果。该方法利用对立色模型提高了色彩特征对光照变化的鲁棒性,并且综合利用彩色图像的色彩和纹理信息提高了人脸识别的精度,特别是对模糊图像的识别精度。在彩色FERET人脸库和AR人脸库上的实验表明,相对于直接对灰度图像进行识别的方法,该方法对清晰图像的识别率提高了4.5%～16.3%,而对模糊图像的识别率提升更加显著。     

基于全局和局部特征融合的图像匹配算法研究

针对移动机器人视觉同时定位与地图构建过程中图像处理速度慢以及特征点匹配实时性和准确性差的问题,提出基于颜色特征和改进SURF算法融合的图像匹配算法。首先,采用颜色特征对图像序列进行粗匹配,选取与测试图像最相近的5幅图像作为待匹配图像;其次,改进SURF算法,用Krawtchouk矩对采用Hessian矩阵获取的关键点进行描述,计算关键点的梯度方向和幅值,得到新的特征向量,对待匹配图像提取改进SURF特征再与测试图像进行精确匹配,得到最佳匹配图像,此匹配算法提高了移动机器人图像处理的速度和精度。实验结果表明,改进算法的误匹配率降低10%左右,程序运行时间减少,在可靠性得到保证的同时适应于实时性应用。