基于背景估计和边缘检测的文档图像二值化

为提高文档图像在对比度低、光照不均、大块噪声等复杂图像背景下二值化效果,提出一种背景估计与边缘检测相结合的图像二值化方法。该方法先利用Sauvola算法有效地估计图像的背景,并在此基础上,结合改进的Canny算法获取边缘变化信息,利用基于局部阈值的策略进行二值化。实验结果表明,该算法取得了较好的二值化效果,在开放的DIBCO 2011数据集上测试,其性能与Otsu、Niblack、Sauvola经典方法相比有大幅提高,其F值比ICDAR2011二值化竞赛中第一名的算法略有提高。     

基于离线参数调整的古籍图像二值化算法

古籍退化对古籍图像二值化带来较大挑战,为此,提出一种基于离线参数调整的古籍图像二值化算法.算法分为两步,估计古籍图像背景局部均值,结合基于拉普拉斯能量的二值化算法对古籍进行二值化;根据迭代弗里德曼竞赛算法设计一种离线参数调整方法优化算法中的参数配置.所提算法在H-DIBCO 2016和H-DIBCO 2018数据集上的...     

背景估计和局部自适应集成的手写图像二值化

手写文档图像中存在光照不均、笔墨浸染、纸张退化、阴影等复杂情况,针对文档图像在复杂背景下二值化后OCR效果不理想的问题,提出了一种对改进的背景估计和局部自适应集成的二值化方法。首先利用局部自适应方法得到具有高召回率的二值化图像,然后对背景估计的方法进行改进得到具有高精确率的二值化图像,最后基于连通域的方法将两种类型的图像集成得到结果。使用4种评价指标在DIBCO2013和DIBCO2016手写数据集上进行了对比实验,结果表明该方法整体性能优于Otsu, Wolf, Niblack, Sauvola, Singh和Howe等经典算法。     

分离复杂背景下的文档图像二值化方法

目的 二值化方法的主要依据是像素的颜色和对比度等低级语义特征，辨别出与文字具有相似低级特征的复杂背景是二值化亟待解决的问题。针对文档图像二值化复杂背景分离问题，提出一种分离文档图像复杂背景的二阶段二值化方法。方法 该方法分为易误判像素筛选和二值化分割两个处理阶段，根据两个阶段的分工构建不同结构的两个网络，前者强化对复杂背景中易误判像素识别和分离能力，后者着重文字像素准确预测，以此提升整个二值化方法在复杂背景图像上的处理效果；两个网络各司其职，可在压缩参数量的前提下出色完成各自任务，进一步提高网络效率。同时，为了增强文字目标细节处理能力，提出一种非对称编码—解码结构，给出两种组合方式。结果 实验在文本图像二值化比赛（competition on document image binarization，DIBCO）的DIBCO2016、DIBCO2017以及DIBCO2018数据集上与其他方法进行比较，本文方法在DIBCO2018中FM（F-measure）为92.35%，仅比经过特殊预处理的方法差0.17%，综合效果均优于其他方法；在DIBCO2017和DIBCO2016中FM分别为93.46%和92.13%，综合效果在所有方法中最好。实验结果表明，非对称编码—解码结构二值化分割的各项指标均有不同程度的提升。结论 提出的二阶段方法能够有效区分复杂背景，进一步提升二值化效果，并在DIBCO数据集上取得了优异成绩。开源代码网址为https://github.com/wjlbnw/Mask_Detail_Net。     

低质量文档图像二值化算法研究

针对低质量文档图像中存在的墨迹浸润、页面污渍或背景纹理等退化因素,提出一种低质量文档图像二值化算法。算法首先基于文档图像的局部对比度实现字符笔画像素检测,然后采用Otsu算法对其进行全局最优阈值化处理,最后通过估计字符笔画宽度确定邻域窗尺寸,从而实现字符前景与页面背景的精细分割。实验结果表明,该算法在F-measure、PSNR、SSIM、NRM、DRD等性能指标方面较其它经典的文档二值化算法具有明显优势。该算法不仅能够较好地保留笔画细节外,还能够较好地抑制文档背景。     

基于支持向量机的低质量文档图像二值化

针对低质量文档图像受墨迹浸润、页面污渍、背景纹理或光照不均等因素的影响,提出一种基于支持向量机(SVM)的低质量文档图像二值化方法。该方法对文档图像进行分块,并增强每个图像块的局部对比度;利用SVM将这些图像块分成三类,对不同图像类采用不同的阈值处理,实现粗略分割;通过笔画宽度估计确定邻域窗尺寸,从而实现局部精细二值化。实验结果表明,该算法无论从二值化图像质量,还是各种评估参数,较其他经典文档二值化方法都具有明显优势。     

一种基于马尔科夫随机场的蒙古文古籍图像恢复方法

针对蒙古文古籍图像检索领域中对同一查询关键词,不同的二值化算法对整体检索性能影响问题,提出一种基于马尔科夫随机场的蒙古文古籍图像二值化方法,从而提高蒙古文古籍图像的检索性能。利用马尔科夫随机场模型在灰度图像和二值图像之间建模,通过训练码本估计隐藏层的先验概率,并分析灰度图像的直方图估计可观察层的概率密度。利用这两种先验知识实现图像二值化。实验数据集为100页蒙古文《甘珠尔经》,为了验证本文所提方法的性能,实验采用R-Precision作为评价指标。实验结果表明,基于马尔科夫随机场的二值化方法不仅可以有效修复受损图像,还可以进一步提高其检索性能。     

一种改进的文本图像二值化方法

针对常用的局部阈值方法-Niblack算法中存在的问题,提出了一种改进的文本图像二值化算法。改进后的Niblack算法对背景灰度不均匀的图像具有良好的适应性,抗噪声能力强,保持笔画连通性好,更适合于文本图像的二值化。实验证明了该算法的有效性。     

一种可抗二值化攻击的文本图像可见水印算法

一些在文本图像中嵌入可见水印标识的方法会在二值化攻击下完全失效,因此提出一种基于灰度均匀分布的文本图像可见水印算法。该算法通过对二值水印图像的黑色像素进行概率筛选来控制水印的嵌入强度,然后将二值文本图像和筛选后的水印图像映射到相同的灰度分布范围,以得到含可见水印标识的文本水印作品。仿真实验表明,该算法生成的文本图像可见水印作品灰度均匀分布,能够抵抗二值化攻击,具有良好的鲁棒性。     

融合背景估计与U-Net的文档图像二值化算法

针对低质量文档图像存在页面污渍、墨迹浸润、背景纹理等多种退化因素,提出一种融合背景估计与U型卷积神经网络（U-Net）的文档图像二值化算法。该算法首先进行图像对比度增强,然后通过形态学闭操作来估计文档图像背景,并利用全卷积网络,即U-Net对背景减除图像进行前景背景分割,最后采用全局最优阈值处理方法获得最终二值图像。实验结果表明,在2016和2017年国际文档图像二值化竞赛中该算法的◢F值（F◣-measure,FM）、伪◢F◣值（pseudo ◢F◣-measure,p-FM）、峰值信噪比（peak signal to noise ratio,PSNR）、距离倒数失真度量（distance reciprocal distortion,DRD）比性能次优的经典算法最高有5.58%、2.47%、0.86 dB、1.19%的性能提升。     

Document image binarization using background estimation and stroke edges

Document images often suffer from different types of degradation that renders the document image binarization a challenging task. This paper presents a document image binarization technique that segments the text from badly degraded document images accurately. The proposed technique is based on the observations that the text documents usually have a document background of the uniform color and texture and the document text within it has a different intensity level compared with the surrounding document background. Given a document image, the proposed technique first estimates a document background surface through an iterative polynomial smoothing procedure. Different types of document degradation are then compensated by using the estimated document background surface. The text stroke edge is further detected from the compensated document image by using L1-norm image gradient. Finally, the document text is segmented by a local threshold that is estimated based on the detected text stroke edges. The proposed technique was submitted to the recent document image binarization contest (DIBCO) held under the framework of ICDAR 2009 and has achieved the top performance among 43 algorithms that are submitted from 35 international research groups.     

基于改进U-Net的低质量文本图像二值化

文本图像二值化是光学字符识别的关键步骤,但低质量文本图像背景噪声复杂,且图像全局上下文信息以及深层抽象信息难以获取,使得最终的二值化结果中文字区域分割不精确、文字的形状和轮廓等特征表达不足,从而导致二值化效果不佳。为此,提出一种基于改进U-Net网络的低质量文本图像二值化方法。采用适合小数据集的分割网络U-Net作为骨干模型,选择预训练的VGG16作为U-Net的编码器以提升模型的特征提取能力。通过融合轻量级全局上下文块的U-Net瓶颈层实现特征图的全局上下文建模。在U-Net解码器的各上采样块中融合残差跳跃连接,以提升模型的特征还原能力。从上述编码器、瓶颈层和解码器3个方面分别对U-Net进行改进,从而实现更精确的文本图像二值化。在DIBCO 2016—2018数据集上的实验结果表明,相较Otsu、Sauvola等方法,该方法能够实现更好的去噪效果,其二值化结果中保留了更多的细节特征,文字的形状和轮廓更精确、清晰。     

Adaptive binarization of severely degraded and non-uniformly illuminated documents

This paper presents a new adaptive binarization method for the degraded document images. Variable background, non-uniform illumination, and blur caused by humidity are the addressed degradations. The proposed method has four steps: contrast analysis, which calculates the local contrast threshold; contrast stretching, thresholding by computing global threshold; and noise removal to improve the quality of binarized image. Evaluation of proposed method has been done using optical character recognition, visual criteria, and established measures: execution time, F-measure, peak signal-to-noise ratio, negative rate metric, and information to noise difference. Our method is tested on the four types of datasets including Document Image Binarization Contest (DIBCO) series datasets (DIBCO 2009, H-DIBCO 2010, and DIBCO 2011), which include a variety of degraded document images. On the basis of evaluation measures, the results of proposed method are promising and achieved good performance after extensive testing with eight techniques referred in the literature.     

A local linear level set method for the binarization of degraded historical document images

Document image binarization is a difficult task, especially for complex document images. Nonuniform background, stains, and variation in the intensity of the printed characters are some examples of challenging document features. In this work, binarization is accomplished by taking advantage of local probabilistic models and of a flexible active contour scheme. More specifically, local linear models are used to estimate both the expected stroke and the background pixel intensities. This information is then used as the main driving force in the propagation of an active contour. In addition, a curvature-based force is used to control the viscosity of the contour and leads to more natural-looking results. The proposed implementation benefits from the level set framework, which is highly successful in other contexts, such as medical image segmentation and road network extraction from satellite images. The validity of the proposed approach is demonstrated on both recent and historical document images of various types and languages. In addition, this method was submitted to the Document Image Binarization Contest (DIBCO??09), at which it placed 3rd.     

An adaptive local binarization method for document images based on a novel thresholding method and dynamic windows

Binary image representation is essential format for document analysis. In general, different available binarization techniques are implemented for different types of binarization problems. The majority of binarization techniques are complex and are compounded from filters and existing operations. However, the few simple thresholding methods available cannot be applied to many binarization problems. In this paper, we propose a local binarization method based on a simple, novel thresholding method with dynamic and flexible windows. The proposed method is tested on selected samples called the DIBCO 2009 benchmark dataset using specialized evaluation techniques for binarization processes. To evaluate the performance of our proposed method, we compared it with the Niblack, Sauvola and NICK methods. The results of the experiments show that the proposed method adapts well to all types of binarization challenges, can deal with higher numbers of binarization problems and boosts the overall performance of the binarization.     

Binarization of degraded document image based on feature space partitioning and classification

In this paper, we propose a new algorithm for the binarization of degraded document images. We map the image into a 2D feature space in which the text and background pixels are separable, and then we partition this feature space into small regions. These regions are labeled as text or background using the result of a basic binarization algorithm applied on the original image. Finally, each pixel of the image is classified as either text or background based on the label of its corresponding region in the feature space. Our algorithm splits the feature space into text and background regions without using any training dataset. In addition, this algorithm does not need any parameter setting by the user and is appropriate for various types of degraded document images. The proposed algorithm demonstrated superior performance against six well-known algorithms on three datasets.     

结构先验指导的文本图像修复模型

目的 图像修复是根据图像中已知内容来自动恢复丢失内容的过程。目前基于深度学习的图像修复模型在自然图像和人脸图像修复上取得了一定效果，但是鲜有对文本图像修复的研究，其中保证结构连贯和纹理一致的方法也没有关注文字本身的修复。针对这一问题，提出了一种结构先验指导的文本图像修复模型。方法 首先以Transformer为基础，构建一个结构先验重建网络，捕捉全局依赖关系重建文本骨架和边缘结构先验图像，然后提出一种新的静态到动态残差模块（static-to-dynamic residual block，StDRB），将静态特征转换到动态文本图像序列特征，并将其融合到编码器—解码器结构的修复网络中，在结构先验指导和梯度先验损失等联合损失的监督下，使修复后的文本笔划连贯，内容真实自然，达到有利于下游文本检测和识别任务的目的。结果 实验在藏文和英文两种语言的合成数据集上，与4种图像修复模型进行了比较。结果表明，本文模型在主观视觉感受上达到了较好的效果，在藏文和英文数据集上的峰值信噪比和结构相似度分别达到了42.31 dB，98.10%和39.23 dB，98.55%，使用Tesseract OCR （optical character recognition）识别修复后藏文图像中的文字的准确率达到了62.83%，使用Tesseract OCR、CRNN （convolutional recurrent neural network）以及ASTER （attentional scene text recognizer）识别修复后英文图像中的文字的准确率分别达到了85.13%，86.04%和76.71%，均优于对比模型。结论 本文提出的文本图像修复模型借鉴了图像修复方法的思想，利用文本图像中文字本身的特性，取得了更加准确的文本图像修复结果。     

Optimal combination of document binarization techniques using a self-organizing map neural network

This paper proposes an integrated system for the binarization of normal and degraded printed documents for the purpose of visualization and recognition of text characters. In degraded documents, where considerable background noise or variation in contrast and illumination exists, there are many pixels that cannot be easily classified as foreground or background pixels. For this reason, it is necessary to perform document binarization by combining and taking into account the results of a set of binarization techniques, especially for document pixels that have high vagueness. The proposed binarization technique takes advantage of the benefits of a set of selected binarization algorithms by combining their results using a Kohonen self-organizing map neural network. Specifically, in the first stage the best parameter values for each independent binarization technique are estimated. In the second stage and in order to take advantage of the binarization information given by the independent techniques, the neural network is fed by the binarization results obtained by those techniques using their estimated best parameter values. This procedure is adaptive because the estimation of the best parameter values depends on the content of images. The proposed binarization technique is extensively tested with a variety of degraded document images. Several experimental and comparative results, exhibiting the performance of the proposed technique, are presented.     

单通道双谱夜视系统中的图像修复及消噪技术

基于单通道双谱夜视系统中对图像修复的需要,提出了一种基于水平集的图像修复及消噪技术。在阐述单通道双谱夜视系统工作原理的基础上,结合系统的实时性要求及待修复微光条纹图像的特点,设计了处理速度高的水平集修复算法;考虑微光图像的噪声对修复结果的影响,在修复的同时加入了热传导滤波方法。实验结果表明,该算法能够对微光条纹图像进行实时且有效的修复。