Scene word recognition from pieces to whole

Convolutional neural networks (CNNs) have had great success with regard to the object classification problem. For character classification, we found that training and testing using accurately segmented character regions with CNNs resulted in higher accuracy than when roughly segmented regions were used. Therefore, we expect to extract complete character regions from scene images. Text in natural scene images has an obvious contrast with its attachments. Many methods attempt to extract characters through different segmentation techniques. However, for blurred, occluded, and complex background cases, those methods may result in adjoined or over segmented characters. In this paper, we propose a scene word recognition model that integrates words from small pieces to entire after-cluster-based segmentation. The segmented connected components are classified as four types: background, individual character proposals, adjoined characters, and stroke proposals. Individual character proposals are directly inputted to a CNN that is trained using accurately segmented character images. The sliding window strategy is applied to adjoined character regions. Stroke proposals are considered as fragments of entire characters whose locations are estimated by a stroke spatial distribution system. Then, the estimated characters from adjoined characters and stroke proposals are classified by a CNN that is trained on roughly segmented character images. Finally, a lexicondriven integration method is performed to obtain the final word recognition results. Compared to other word recognition methods, our method achieves a comparable performance on Street View Text and the ICDAR 2003 and ICDAR 2013 benchmark databases. Moreover, our method can deal with recognizing text images of occlusion and improperly segmented text images.     

Segmentation and labeling of face images for electronic documents

Face image segmentation and labeling is required in several quality tests which a face image has to pass in order to be included into an electronic ID document. The complexity of such a problem depends on the complexity of the scene, but in general there are no restrictions to the scene. The procedure that we have developed segments a face image into five regions: skin, hair, shoulders, background and padding frame. The presented method consists of two main steps: oversegmentation and labeling. In the first step, the image is segmented into homogeneous regions, whereas in the second step, the labeling of the homogeneous regions is performed. In the course of our research we experimented with several methods for the two described steps, and in this paper we present a setup in which the oversegmentation is performed using the mean-shift segmentation, and labeling is performed using the AdaBoost classification algorithm. Such setup has produced the best results in our experiments which we also present herein.     

结合超体素和区域增长的植物器官点云分割

点云分割是点云识别与建模的基础。为提高点云分割准确率和效率，提出一种结合超体素和区域增长的自适应分割算法。根据三维点云的空间位置和法向量信息，利用八叉树对点云进行初始分割得到超体素。选取超体素的中心体素组成一个新的重采样后的密度均匀点云，降低原始点云数据处理量，从而减少运算时间。建立重采样后点云数据的K-D树索引，根据其局部特征得到点云簇。最后将聚类结果返回到原始点云空间。分别选取植物三个物候期的激光扫描点云，对该方法的有效性进行验证。实验结果表明，该方法分割后点云与手工分割平均拟合度达到93.38%，高于其他同类方法，且算法效率得到明显提升。     

Interactive image segmentation via kernel propagation

In this paper, we propose a new approach to interactive image segmentation via kernel propagation (KP), called KP Cut. The key to success in interactive image segmentation is to preserve characteristics of the user?s interactive input and maintain data-coherence effectively. To achieve this, we employ KP which is very effective in propagating the given supervised information into the entire data set. KP first learns a small-size seed-kernel matrix, and then propagates it into a large-size full-kernel matrix. It is based on a learned kernel, and thus can fit the given data better than a predefined kernel. Based on KP, we first generate a small-size seed-kernel matrix from the user?s interactive input. Then, the seed-kernel matrix is propagated into the full-kernel matrix of the entire image. During the propagation, foreground objects are effectively segmented from background. Experimental results demonstrate that KP Cut effectively extracts foreground objects from background, and outperforms the state-of-the-art methods for interactive image segmentation.     

ECISER: Efficient Clip-art Image SEgmentation by Re-rasterization

Clip-art image segmentation is widely used as an essential step to solve many vision problems such as colorization and vectorization. Many of these applications not only demand accurate segmentation results, but also have little tolerance for time cost, which leads to the main challenge of this kind of segmentation. However, most existing segmentation techniques are found not sufficient for this purpose due to either their high computation cost or low accuracy. To address such issues, we propose a novel segmentation approach, ECISER, which is well-suited in this context. The basic idea of ECISER is to take advantage of the particular nature of cartoon images and connect image segmentation with aliased rasterization. Based on such relationship, a clip-art image can be quickly segmented into regions by re-rasterization of the original image and several other computationally efficient techniques developed in this paper. Experimental results show that our method achieves dramatic computational speedups over the current state-of-the-art approaches, while preserving almost the same quality of results.     

Error measures for scene segmentation

Scene segmentation is an important problem in pattern recognition. Current subjective methods for evaluation and comparison of scene segmentation techniques are inadequate and objective quantitative measures are desirable. Two error measures, the percentage area misclassified (p) and a new pixel distance error (ε) were defined and evaluated in terms of their correlation with human observation for comparison of multiple segmentations of the same scene and multiple scenes segmented by the same technique. The results indicate that both these measures can be helpful in the evaluation and comparison of scene segmentation procedures.     

Image Segmentation Using Some Piecewise Constant Level Set Methods with MBO Type of Projection

In this work, we are trying to propose fast algorithms for Mumford-Shah image segmentation using some recently proposed piecewise constant level set methods (PCLSM). Two variants of the PCLSM will be considered in this work. The first variant, which we call the binary level set method, needs a level set function which only takes values ±1 to identify the regions. The second variant only needs to use one piecewise constant level set function to identify arbitrary number of regions. For the Mumford-Shah image segmentation model with these new level set methods, one needs to minimize some smooth energy functionals under some constrains. A penalty method will be used to deal with the constraint. AOS (additive operator splitting) and MOS (multiplicative operator splitting) schemes will be used to solve the Euler-Lagrange equations for the minimization problems. By doing this, we obtain some algorithms which are essentially applying the MBO scheme for our segmentation models. Advantages and disadvantages are discussed for the proposed schemes. We acknowledge support from the Norwegian Research Council and IMS of the National University of Singapore.     

DREAM2S: Deformable Regions Driven by an Eulerian Accurate Minimization Method for Image and Video Segmentation

This paper deals with image and video segmentation using active contours. We propose a general form for the energy functional related to region-based active contours. We compute the associated evolution equation using shape derivation tools and accounting for the evolving region-based terms. Then we apply this general framework to compute the evolution equation from functionals that include various statistical measures of homogeneity for the region to be segmented. Experimental results show that the determinant of the covariance matrix appears to be a very relevant tool for segmentation of homogeneous color regions. As an example, it has been successfully applied to face segmentation in real video sequences.     

A novel method for pulmonary embolism detection in CTA images

In this paper, we propose a new computer-aided detection (CAD) – based method to detect pulmonary embolism (PE) in computed tomography angiography images (CTAI). Since lung vessel segmentation is the main objective to provide high sensitivity in PE detection, this method performs accurate lung vessel segmentation. To concatenate clogged vessels due to PEs, the starting region of PEs and some reference points (RPs) are determined. These RPs are detected according to the fixed anatomical structures. After lung vessel tree is segmented, the region, intensity, and size of PEs are used to distinguish them. We used the data sets that have heart disease or abnormal tissues because of lung disease except PE in this work. According to the results, 428 of 450 PEs, labeled by the radiologists from 33 patients, have been detected. The sensitivity of the developed system is 95.1% at 14.4 false positive per data set (FP/ds). With this performance, the proposed CAD system is found quite useful to use as a second reader by the radiologists.     

一种基于水平集的图像快速多区域分割方法

提出了一种新的基于水平集的图像快速多区域分割方法。首先,在经典水平集分割算法的基础上,通过使用新的水平集初始化函数,有效地改善了水平集分割算法的时间性能;其次,通过引入区域分割控制条件控制水平集函数的收敛过程,实现多区域分割。实验结果表明,提出的多区域分割方法具有较好的分割性能,并且时间耗费少。     

3D Graph cut with new edge weights for cerebral white matter segmentation

Accurate and efficient automatic or semi-automatic brain image segmentation methods are of great interest to both scientific and clinical researchers of the human central neural system. Cerebral white matter segmentation in brain Magnetic Resonance Imaging (MRI) data becomes a challenging problem due to a combination of several factors like low contrast, presence of noise and imaging artifacts, partial volume effects, intrinsic tissue variation due to neurodevelopment and neuropathologies, and the highly convoluted geometry of the cortex. In this paper, we propose a new set of edge weights for the traditional graph cut algorithm (Boykov and Jolly, 2001) to correctly segment the cerebral white matter from T1-weighted MRI sequence. In this algorithm, the edge weights of Boykov and Jolly (2001) are modified by comparing the probabilities of an individual voxel and its neighboring voxels to belong to different segmentation classes. A shape prior in form of a series of ellipses is used next to model the contours of the human skull in various 2D slices in the sequence. This shape constraint is imposed to prune the original graph constructed from the input to form a subgraph consisting of voxels within the skull contours. Our graph cut algorithm with new set of edge weights is applied to the above subgraph, thereby increasing the segmentation accuracy as well as decreasing the computation time. Average segmentation errors for the proposed algorithm, the graph cut algorithm (Boykov and Jolly, 2001), and the Expectation Maximization Segmentation (EMS) algorithm Van Leemput et al., 2001 in terms of Dice coefficients are found to be (3.72 ± 1.12)%, (14.88 ± 1.69)%, and (11.95 ± 5.2)%, respectively.     

Automatic mass segmentation on mammograms combining random walks and active contour

Accurate mass segmentation on mammograms is a critical step in computer-aided diagnosis (CAD) systems. It is also a challenging task since some of the mass lesions are embedded in normal tissues and possess poor contrast or ambiguous margins. Besides, the shapes and densities of masses in mammograms are various. In this paper, a hybrid method combining a random walks algorithm and Chan-Vese (CV) active contour is proposed for automatic mass segmentation on mammograms. The data set used in this study consists of 1095 mass regions of interest (ROIs). First, the original ROI is preprocessed to suppress noise and surrounding tissues. Based on the preprocessed ROI, a set of seed points is generated for initial random walks segmentation. Afterward, an initial contour of mass and two probability matrices are produced by the initial random walks segmentation. These two probability matrices are used to modify the energy function of the CV model for prevention of contour leaking. Lastly, the final segmentation result is derived by the modified CV model, during which the probability matrices are updated by inserting several rounds of random walks. The proposed method is tested and compared with other four methods. The segmentation results are evaluated based on four evaluation metrics. Experimental results indicate that the proposed method produces more accurate mass segmentation results than the other four methods.     

Seg-CapNet:A Capsule-Based Neural Network for the Segmentation of Left Ventricle from Cardiac Magnetic Resonance Imaging

Deep neural networks (DNNs) have been extensively studied in medical image segmentation.However,existing DNNs often need to train shape models for each object to be segmented,which may yield results that violate cardiac anatomical structure when segmenting cardiac magnetic resonance imaging (MRI).In this paper,we propose a capsule-based neural network,named Seg-CapNet,to model multiple regions simultaneously within a single training process.The Seg-CapNet model consists of the encoder and the decoder.The encoder transforms the input image into feature vectors that represent objects to be segmented by convolutional layers,capsule layers,and fully-connected layers.And the decoder transforms the feature vectors into segmentation masks by up-sampling.Feature maps of each down-sampling layer in the encoder are connected to the corresponding up-sampling layers,which are conducive to the backpropagation of the model.The output vectors of Seg-CapNet contain low-level image features such as grayscale and texture,as well as semantic features including the position and size of the objects,which is beneficial for improving the segmentation accuracy.The proposed model is validated on the open dataset of the Automated Cardiac Diagnosis Challenge 2017 (ACDC 2017) and the Sunnybrook Cardiac Magnetic Resonance Imaging (MRI) segmentation challenge.Experimental results show that the mean Dice coefficient of Seg-CapNet is increased by 4.7％ and the average Hausdorff distance is reduced by 22％.The proposed model also reduces the model parameters and improves the training speed while obtaining the accurate segmentation of multiple regions.     

Seg-CapNet:A Capsule-Based Neural Network for the Segmentation of Left Ventricle from Cardiac Magnetic Resonance Imaging

Deep neural networks (DNNs) have been extensively studied in medical image segmentation.However,existing DNNs often need to train shape models for each object to be segmented,which may yield results that violate cardiac anatomical structure when segmenting cardiac magnetic resonance imaging (MRI).In this paper,we propose a capsule-based neural network,named Seg-CapNet,to model multiple regions simultaneously within a single training process.The Seg-CapNet model consists of the encoder and the decoder.The encoder transforms the input image into feature vectors that represent objects to be segmented by convolutional layers,capsule layers,and fully-connected layers.And the decoder transforms the feature vectors into segmentation masks by up-sampling.Feature maps of each down-sampling layer in the encoder are connected to the corresponding up-sampling layers,which are conducive to the backpropagation of the model.The output vectors of Seg-CapNet contain low-level image features such as grayscale and texture,as well as semantic features including the position and size of the objects,which is beneficial for improving the segmentation accuracy.The proposed model is validated on the open dataset of the Automated Cardiac Diagnosis Challenge 2017 (ACDC 2017) and the Sunnybrook Cardiac Magnetic Resonance Imaging (MRI) segmentation challenge.Experimental results show that the mean Dice coefficient of Seg-CapNet is increased by 4.7％ and the average Hausdorff distance is reduced by 22％.The proposed model also reduces the model parameters and improves the training speed while obtaining the accurate segmentation of multiple regions.     

基于多层水平集函数的三维多相图像分割

多相图像分割通常利用多个水平集函数分别定义不同区域的特征函数,其极值求解问题需要对多个函数分别求极值,计算效率较低。针对三维多相图像,提出一种改进的变分水平集模型,采用一个多层水平集函数的n层水平集隐式曲面,将图像划分为n个区域,通过对一个水平集函数求极值,实现三维多相分段常值图像的快速分割与重建。将能量泛函表达为数据项和规则项,借助规则化Heaviside函数设计区域划分的通用特征函数,采用Split-Bregman投影方法进行能量最小化求解。实验结果表明,该模型可以有效地实现三维多相图像分割,与Chan-Vese模型相比,其迭代步数较少,分割速度较快。     

Watershed from propagated markers: An interactive method to morphological object segmentation in image sequences

In this paper, it is introduced an interactive method to object segmentation in image sequences, by combining classical morphological segmentation with motion estimation – the watershed from propagated markers. In this method, the objects are segmented interactively in the first frame and the mask generated by its segmentation provides the markers that will be used to track and segment the object in the next frame. Besides the interactivity, the proposed method has the following important characteristics: generality, rapid response and progressive manual edition. This paper also introduces a new benchmark to do quantitative evaluation of assisted object segmentation methods applied to image sequences. The evaluation is done according to several criteria such as the robustness of segmentation and the easiness to segment the objects through the sequence.     

Image segmentation from scale and rotation invariant texture features from the double dyadic dual-tree complex wavelet transform

A goal of image segmentation is to divide an image into regions that have some semantic meaning. Because regions of semantic meaning often include variations in colour and intensity, various segmentation algorithms that use multi-pixel textures have been developed. A challenge for these algorithms is to incorporate invariance to rotation and changes in scale. In this paper, we propose a new scale and rotation invariant, texture-based segmentation algorithm, that performs feature extraction using the Dual-Tree Complex Wavelet Transform (DT-CWT). The DT-CWT is used to analyse a signal at, and between, dyadic scales. The performance of image segmentation using this new method is compared with existing techniques over different imagery databases with operator produced ground truth data. Compared with previous algorithms, our segmentation results show that the new texture feature is capable of performing well over general images and particularly well over images containing objects with scaled and rotated textures.     

分区域去运动模糊

去运动模糊一直是计算机视觉领域中面向画质增强的一个热点研究方向。模糊核的估算是去运动模糊中的关键问题。提出一种新的思路,即首先将模糊图像按照模糊核的相似度进行图像分割,再对分割后的图像应用空间不变去模糊算法。本文方法主要包含以下几个步骤:分离输入图像中的光照、颜色和纹理信息;分割图像;分区域估算模糊核,计算重叠区域模糊核,并根据计算出的模糊核进行分区域单核去模糊;利用重叠区域整合拼接去模糊结果并还原光照和颜色信息。实验结果表明本文方法比基于单核的去运动模糊算法效果要好。     

基于边界特征的三维模型分割

点云分割是三维模型检索、分类及重建的基础,为解决点云分割算法存在鲁棒性差、过分割和欠分割问题,提出一种基于边界特征的点云模型分割算法。将点云模型过分割为弱凸区域,利用巴氏距离判断相邻区域的相似性进行区域合并,采用改进的形状直径函数进行最终合并。由主流评价方法及实验证明,大多数模型可以取得良好的分割效果。     

中文分词模型词典融入方法比较

目前比较流行的中文分词方法为基于统计模型的机器学习方法。基于统计的方法一般采用人工标注的句子级的标注语料进行训练,但是这种方法往往忽略了已有的经过多年积累的人工标注的词典信息。这些信息尤其是在面向跨领域时,由于目标领域句子级别的标注资源稀少,从而显得更加珍贵。因此如何充分而且有效的在基于统计的模型中利用词典信息,是一个非常值得关注的工作。最近已有部分工作对它进行了研究,按照词典信息融入方式大致可以分为两类：一类是在基于字的序列标注模型中融入词典特征,而另一类是在基于词的柱搜索模型中融入特征。对这两类方法进行比较,并进一步进行结合。实验表明,这两类方法结合之后,词典信息可以得到更充分的利用,最终无论是在同领域测试和还是在跨领域测试上都取得了更优的性能。