Eikonal-based region growing for efficient clustering

We describe an Eikonal-based algorithm for computing dense oversegmentation of an image, often called superpixels. This oversegmentation respects local image boundaries while limiting undersegmentation. The proposed algorithm relies on a region growing scheme, where the potential map used is not fixed and evolves during the diffusion. Refinement steps are also proposed to enhance at low cost the first oversegmentation. Quantitative comparisons on the Berkeley dataset show good performance on traditional metrics over current state-of-the art superpixel methods.     

An review on superpixels generation algorithms

Superpixel segmentation is the oversegmentation of an image into a set of homogeneous regions. Superpixel has many specific properties and has been commonly used as supporting regions for primitives to reduce computations in various computer vision tasks. One property of superpixels is compactness, which is preferred in some applications. In this paper, we give an review on image superpixel segmentation algorithms proposed in recent years. Superpixel segmentation approaches are classified based on the compactness constraint and their main idea are introduced. We also compare these algorithms in visual and evaluate them with five common measurements.     

Fast and Scalable Mesh Superfacets

In the field of computer vision, the introduction of a low‐level preprocessing step to oversegment images into superpixels – relatively small regions whose boundaries agree with those of the semantic entities in the scene – has enabled advances in segmentation by reducing the number of elements to be labeled from hundreds of thousands, or millions, to a just few hundred. While some recent works in mesh processing have used an analogous oversegmentation, they were not intended to be general and have relied on graph cut techniques that do not scale to current mesh sizes. Here, we present an iterative superfacet algorithm and introduce adaptations of undersegmentation error and compactness, which are well‐motivated and principled metrics from the vision community. We demonstrate that our approach produces results comparable to those of the normalized cuts algorithm when evaluated on the Princeton Segmentation Benchmark, while requiring orders of magnitude less time and memory and easily scaling to, and enabling the processing of, much larger meshes.     

Automatic recognition of handwritten numerical strings: a recognition and verification strategy

A modular system to recognize handwritten numerical strings is proposed. It uses a segmentation-based recognition approach and a recognition and verification strategy. The approach combines the outputs from different levels such as segmentation, recognition, and postprocessing in a probabilistic model. A new verification scheme which contains two verifiers to deal with the problems of oversegmentation and undersegmentation is presented. A new feature set is also introduced to feed the oversegmentation verifier. A postprocessor based on a deterministic automaton is used and the global decision module makes an accept/reject decision. Finally, experimental results on two databases are presented: numerical amounts on Brazilian bank checks and NIST SD19. The latter aims at validating the concept of modular system and showing the robustness of the system using a well-known database.     

融合T节点线索的图像物体分割

目的 图像分割是计算机视觉、数字图像处理等应用领域首要解决的关键问题。针对现有的单幅图像物体分割算法广泛存在的过分割和过合并现象,提出基于图像T型节点线索的图像物体分割算法。方法 首先,利用L0梯度最小化方法平滑目标图像,剔除细小纹理的干扰;其次,基于Graph-based分割算法对平滑后图像进行适度分割,得到粗糙分割结果;最后,借助于图像中广泛存在的T型节点线索对初始分割块进行区域合并得到最终优化分割结果。结果 将本文算法分别与Grabcut算法及Graph-based算法在不同场景类型下进行了实验与对比。实验结果显示,Grabcut算法需要人工定位边界且一次只能分割单个物体,Graph-based算法综合类内相似度和类间差异性,可以有效保持图像边界,但无法有效控制分割块数量,且分割结果对阈值参数过分依赖,极易导致过分割和过合并现象。本文方法在降低过分割和过合并现象、边界定位精确性和分割准确率方面获得明显改进,几组不同类型的图片分割准确率平均值达到91.16%,明显由于其他算法。处理图像尺寸800×600像素的图像平均耗时3.5 s,较之其他算法略有增加。结论 与各种算法对比结果表明,该算法可有效解决过分割和过合并问题,对比实验结果验证了该方法的有效性,能够取得具有一定语义的图像物体分割结果。     

基于3D直方图的彩色图象分割方法

以阈值分割技术为基础，提出了基于3D直方图的生长法和尺度空间聚类方法。分别利用3D空间中同一类目标像素频度的连通性的尺度空间聚类原理，简单而有效地解决了多维阈值分割带来的过度分割问题。该方法对不同类型图象具有很好的适应性，并已成功应用于医学彩色图象处理领域。     

Color segmentation for text extraction

The capability of extracting and recognizing characters printed in color documents will widen immensely the applications of OCR systems. This paper describes a new method of color segmentation to extract character areas from a color document. At first glance, the characters seem to be printed in a single color, but actual measurements reveal that the color image has a distribution of components. Compared with clustering algorithms, our method prevents oversegmentation and fusion with the background while maintaining real-time usability. It extracts the representative colors based on a histogram analysis of the color space. Our method also contains a selective local color averaging technique that removes the problem of mesh noise on high-resolution color images.Received: 25 July 2003, Revised: 10 August 2003, Published online: 6 February 2004Correspondence to: Hiroyuki Hase. Current address: 3-9-1 Bunkyo, Fukui-shi 910-8507, Japan     

Shape and the Stereo Correspondence Problem

We examine the implications of shape on the process of finding dense correspondence and half-occlusions for a stereo pair of images. The desired property of the disparity map is that it should be a piecewise continuous function which is consistent with the images and which has the minimum number of discontinuities. To zeroth order, piecewise continuity becomes piecewise constancy. Using this approximation, we first discuss an approach for dealing with such a fronto-parallel shapeless world, and the problems involved therein. We then introduce horizontal and vertical slant to create a first order approximation to piecewise continuity. In particular, we emphasize the following geometric fact: a horizontally slanted surface (i.e., having depth variation in the direction of the separation of the two cameras) will appear horizontally stretched in one image as compared to the other image. Thus, while corresponding two images, N pixels on a scanline in one image may correspond to a different number of pixels M in the other image. This leads to three important modifications to existing stereo algorithms: (a) due to unequal sampling, existing intensity matching metrics must be modified, (b) unequal numbers of pixels in the two images must be allowed to correspond to each other, and (c) the uniqueness constraint, which is often used for detecting occlusions, must be changed to an interval uniqueness constraint. We also discuss the asymmetry between vertical and horizontal slant, and the central role of non-horizontal edges in the context of vertical slant. Using experiments, we discuss cases where existing algorithms fail, and how the incorporation of these new constraints provides correct results.     

克服Watershed算法过度分割的方法

Watershed算法是一种形态学的图像分割算法，但由于其对噪声十分敏感，分割结果往往存在过度分割的现象。该文提出一种基于小波分析的Watershed算法，充分利用小波的多分辨率特性有效地解决了Watershed算法的过度分割问题，并大大提高了分割计算速度。     

Skew detection in document images based on rectangular active contour

The digitalization processes of documents produce frequently images with small rotation angles. The skew angles in document images degrade the performance of optical character recognition (OCR) tools. Therefore, skew detection of document images plays an important role in automatic document analysis systems. In this paper, we propose a Rectangular Active Contour Model (RAC Model) for content region detection and skew angle calculation by imposing a rectangular shape constraint on the zero-level set in Chan–Vese Model (C-V Model) according to the rectangular feature of content regions in document images. Our algorithm differs from other skew detection methods in that it does not rely on local image features. Instead, it uses global image features and shape constraint to obtain a strong robustness in detecting skew angles of document images. We experimented on different types of document images. Comparing the results with other skew detection algorithms, our algorithm is more accurate in detecting the skews of the complex document images with different fonts, tables, illustrations, and layouts. We do not need to pre-process the original image, even if it is noisy, and at the same time the rectangular content region of a document image is also detected.     

图像特征匹配中一种高效的鲁棒估计算法

在图像特征匹配过程中,误匹配不可避免。提出一种新的基于拓扑约束（顺序约束和仿射不变约束）的外点去除算法,用于快速地去除图像粗匹配结果中的误配点。该算法 对随机采样集进行拓扑过滤,只对满足拓扑约束的采样集进行计算。实验表明,该算法相比于传统的鲁棒估计算法RANSAC和改进的PROSAC算法,大大提高了计算效率并保持很高的 计算精度,有助于提升图像匹配性能及3维重建的精度和鲁棒性。     

Using In-frame Shear Constraints for Monocular Motion Segmentation of Rigid Bodies

It is a well known result in the vision literature that the motion of independently moving objects viewed by an affine camera lie on affine subspaces of dimension four or less. As a result a large number of the recently proposed motion segmentation algorithms model the problem as one of clustering the trajectory data to its corresponding affine subspace. While these algorithms are elegant in formulation and achieve near perfect results on benchmark datasets, they fail to address certain very key real-world challenges, including perspective effects and motion degeneracies. Within a robotics and autonomous vehicle setting, the relative configuration of the robot and moving object will frequently be degenerate leading to a failure of subspace clustering algorithms. On the other hand, while gestalt-inspired motion similarity algorithms have been used for motion segmentation, in the moving camera case, they tend to over-segment or under-segment the scene based on their parameter values. In this paper we present a principled approach that incorporates the strengths of both approaches into a cohesive motion segmentation algorithm capable of dealing with the degenerate cases, where camera motion follows that of the moving object. We first generate a set of prospective motion models for the various moving and stationary objects in the video sequence by a RANSAC-like procedure. Then, we incorporate affine and long-term gestalt-inspired motion similarity constraints, into a multi-label Markov Random Field (MRF). Its inference leads to an over-segmentation, where each label belongs to a particular moving object or the background. This is followed by a model selection step where we merge clusters based on a novel motion coherence constraint, we call in-frame shear, that tracks the in-frame change in orientation and distance between the clusters, leading to the final segmentation. This oversegmentation is deliberate and necessary, allowing us to assess the relative motion between the motion models which we believe to be essential in dealing with degenerate motion scenarios.We present results on the Hopkins-155 benchmark motion segmentation dataset [27], as well as several on-road scenes where camera and object motion are near identical. We show that our algorithm is competitive with the state-of-the-art algorithms on [27] and exceeds them substantially on the more realistic on-road sequences.     

Fuzzy c-means clustering with non local spatial information for noisy image segmentation

As an effective image segmentation method, the standard fuzzy c-means (FCM) clustering algorithm is very sensitive to noise in images. Several modified FCM algorithms, using local spatial information, can overcome this problem to some degree. However, when the noise level in the image is high, these algorithms still cannot obtain satisfactory segmentation performance. In this paper, we introduce a non local spatial constraint term into the objective function of FCM and propose a fuzzy cmeans clustering algorithm with non local spatial information (FCM_NLS). FCM_NLS can deal more effectively with the image noise and preserve geometrical edges in the image. Performance evaluation experiments on synthetic and real images, especially magnetic resonance (MR) images, show that FCM_NLS is more robust than both the standard FCM and the modified FCM algorithms using local spatial information for noisy image segmentation.     

Infrared Polarization and Intensity Image Fusion Algorithm Based on the Feature Transfer

The features of infrared polarization and intensity images are not finely transferred to the fused image by using traditional fusion algorithms, which leads to a severe blur of the fused image. This study proposes a new infrared polarization and intensity image fusion algorithm based on the feature transfer. First, the contrast features of the infrared polarization image are extracted by the multiscale average filter decomposition with help of standard deviation constraint. The texture features of infrared polarization images are retrieved via non-subsample-shearlet transform at the same time. Second, the difference of the features is measured using the similarity index, which is used as the transfer weight for the infrared polarization feature images during the later phase of the image fusion. Finally, the fused image is obtained by the superimposition of the infrared intensity image and feature images, which are created from the infrared polarization image. The experimental results demonstrated that the proposed method is able to transfer the features of both the infrared intensity image and the polarization image into the fused images. It performs well on both subjective and objective image quality.     

基于自注意力生成对抗网络的图像超分辨率重建

针对如何恢复重建后超分辨率图像的纹理细节问题,提出基于自注意力生成对抗网络的图像超分辨率重建模型(SRAGAN).在SRAGAN中,基于自注意力机制和残差模块相结合的生成器用于将低分辨率图像变换为超分辨率图像,基于深度卷积网络构成的判别器试图区分重建后的超分辨率图像和真实超分辨率图像间的差异.在损失函数构造方面,一方面利用Charbonnier内容损失函数来提高图像的重建精度,另一方面使用预训练VGG网络激活前的特征值来计算感知损失以实现超分辨率图像的精确纹理细节重构.实验结果表明,SRAGAN在峰值信噪比和结构相似度分数上均优于当前流行算法,能够重构出更为真实和具有清晰纹理的图像.     

Watershed segmentation using prior shape and appearance knowledge

Watershed transformation is a common technique for image segmentation. However, its use for automatic medical image segmentation has been limited particularly due to oversegmentation and sensitivity to noise. Employing prior shape knowledge has demonstrated robust improvements to medical image segmentation algorithms. We propose a novel method for enhancing watershed segmentation by utilizing prior shape and appearance knowledge. Our method iteratively aligns a shape histogram with the result of an improved k-means clustering algorithm of the watershed segments. Quantitative validation of magnetic resonance imaging segmentation results supports the robust nature of our method.     

基于自监督约束的双尺度真实图像盲去噪算法

现存的图像去噪算法在处理加性高斯白噪声上已经取得令人满意的效果,然而其在未知噪声强度的真实噪声图像上泛化性能较差.鉴于深度卷积神经网络极大地促进了图像盲去噪技术的发展,针对真实噪声图像提出一种基于自监督约束的双尺度真实图像盲去噪算法.首先,所提算法借助小尺度网络分支得到的初步去噪结果为大尺度分支的图像去噪提供额外的有用信息,以帮助后者实现良好的去噪效果.其次,用于去噪的网络模型由噪声估计子网络和图像非盲去噪子网络构成,其中噪声估计子网络用于预测输入图像的噪声强度,非盲去噪子网络则在所预测的噪声强度指导下进行图像去噪.鉴于真实噪声图像通常缺少对应的清晰图像作为标签,提出了一种基于全变分先验的边缘保持自监督约束和一个基于图像背景一致性的背景自监督约束,前者可通过调节平滑参数将网络泛化到不同的真实噪声数据集上并取得良好的无监督去噪效果,后者则可借助多尺度高斯模糊图像之间的差异信息辅助双尺度网络完成去噪.此外,还提出一种新颖的结构相似性注意力机制,用于引导网络关注图像中微小的结构细节,以便复原出纹理细节更加清晰的真实去噪图像.相关实验结果表明在SIDD,DND和Nam这3个真实基准数据集上,所提的基于自监督的双尺度盲去噪算法无论在视觉效果上还是在量化指标上均优于多种有监督图像去噪方法,且泛化性能也得到了较为明显的提升.     

Synergy between Object Recognition and Image Segmentation Using the Expectation-Maximization Algorithm

In this work, we formulate the interaction between image segmentation and object recognition in the framework of the Expectation-Maximization (EM) algorithm. We consider segmentation as the assignment of image observations to object hypotheses and phrase it as the E-step, while the M-step amounts to fitting the object models to the observations. These two tasks are performed iteratively, thereby simultaneously segmenting an image and reconstructing it in terms of objects. We model objects using Active Appearance Models (AAMs) as they capture both shape and appearance variation. During the E-step, the fidelity of the AAM predictions to the image is used to decide about assigning observations to the object. For this, we propose two top-down segmentation algorithms. The first starts with an oversegmentation of the image and then softly assigns image segments to objects, as in the common setting of EM. The second uses curve evolution to minimize a criterion derived from the variational interpretation of EM and introduces AAMs as shape priors. For the M-step, we derive AAM fitting equations that accommodate segmentation information, thereby allowing for the automated treatment of occlusions. Apart from top-down segmentation results, we provide systematic experiments on object detection that validate the merits of our joint segmentation and recognition approach.     

结合类内和类间距离的可能聚类分割算法

目的 为了进一步提高噪声图像分割的抗噪性和准确性,提出一种结合类内距离和类间距离的改进可能聚类算法并将其应用于图像分割。方法 该算法避免了传统可能性聚类分割算法中仅仅考虑以样本点到聚类中心的距离作为算法的测度,将类内距离与类间距离相结合作为算法的新测度,即考虑了类内紧密程度又考虑了类间离散程度,以便对不同的聚类结构有较强的稳定性和更好的抗噪能力,并且将直方图融入可能模糊聚类分割算法中提出快速可能模糊聚类分割算法,使其对各种较复杂图像的分割具有即时性。结果 通过人工合成图像和实际遥感图像分割测试结果表明,本文改进可能聚类算法是有效的,其分割轮廓清晰,分类准确且噪声较小,其误分率相比其他算法至少降低了2个百分点,同时能获得更满意的分割效果。结论 针对模糊C-均值聚类分割算法和可能性聚类分割算法对于背景和目标颜色相近的图像分类不准确的缺陷,将类内距离与类间距离相结合作为算法的测度有效的解决了图像分割归类问题,并且结合直方图提出快速可能模糊聚类分割算法使其对于大篇幅复杂图像也具有适用性。     

Limits on super-resolution and how to break them

Nearly all super-resolution algorithms are based on the fundamental constraints that the super-resolution image should generate low resolution input images when appropriately warped and down-sampled to model the image formation process. (These reconstruction constraints are normally combined with some form of smoothness prior to regularize their solution.) We derive a sequence of analytical results which show that the reconstruction constraints provide less and less useful information as the magnification factor increases. We also validate these results empirically and show that, for large enough magnification factors, any smoothness prior leads to overly smooth results with very little high-frequency content. Next, we propose a super-resolution algorithm that uses a different kind of constraint in addition to the reconstruction constraints. The algorithm attempts to recognize local features in the low-resolution images and then enhances their resolution in an appropriate manner. We call such a super-resolution algorithm a hallucination or reconstruction algorithm. We tried our hallucination algorithm on two different data sets, frontal images of faces and printed Roman text. We obtained significantly better results than existing reconstruction-based algorithms, both qualitatively and in terms of RMS pixel error