Search | Engineering

订阅投稿

首页工程期刊工程焦点工程成就工程前沿关于我们 English

资源类型

期刊论文 137

会议信息 2

会议视频 2

年份

2023 10

2022 21

2021 9

2020 17

2019 20

2018 10

2017 9

2016 8

2015 7

2014 3

2013 1

2012 4

2011 1

2009 2

2008 2

2007 4

2006 5

2005 1

2004 2

2003 2

展开︾

关键词

人工智能 2

图像分析 2

图像处理 2

深度学习 2

高速铁路 2

AF/PSTM 1

CCD影像 1

PCB缺陷检测 1

PSTM图像分解 1

PSTM数值模拟 1

PSTM消假像 1

Quantitative structure 1

V-W-Mo-Cu催化剂 1

imge analysis 1

k-最近邻分类 1

stereology 1

上下图模型 1

主成分分析 1

展开︾

检索范围：

排序：展示方式：

Optimal CNN-based semantic segmentation model of cutting slope images

Mansheng LIN; Shuai TENG; Gongfa CHEN; Jianbing LV; Zhongyu HAO

《结构与土木工程前沿（英文）》 2022年第16卷第4期页码 414-433 doi: 10.1007/s11709-021-0797-6

摘要： This paper utilizes three popular semantic segmentation networks, specifically DeepLab v3+, fully convolutional network (FCN), and U-Net to qualitively analyze and identify the key components of cutting slope images in complex scenes and achieve rapid image-based slope detection. The elements of cutting slope images are divided into 7 categories. In order to determine the best algorithm for pixel level classification of cutting slope images, the networks are compared from three aspects: a) different neural networks, b) different feature extractors, and c) 2 different optimization algorithms. It is found that DeepLab v3+ with Resnet18 and Sgdm performs best, FCN 32s with Sgdm takes the second, and U-Net with Adam ranks third. This paper also analyzes the segmentation strategies of the three networks in terms of feature map visualization. Results show that the contour generated by DeepLab v3+ (combined with Resnet18 and Sgdm) is closest to the ground truth, while the resulting contour of U-Net (combined with Adam) is closest to the input images.

关键词： slope damage image recognition semantic segmentation feature map visualizations

HTML PDF 收藏

Detecting large-scale underwater cracks based on remote operated vehicle and graph convolutional neural network

Wenxuan CAO; Junjie LI

《结构与土木工程前沿（英文）》 2022年第16卷第11期页码 1378-1396 doi: 10.1007/s11709-022-0855-8

摘要： It is of great significance to quickly detect underwater cracks as they can seriously threaten the safety of underwater structures. Research to date has mainly focused on the detection of above-water-level cracks and hasn’t considered the large scale cracks. In this paper, a large-scale underwater crack examination method is proposed based on image stitching and segmentation. In addition, a purpose of this paper is to design a new convolution method to segment underwater images. An improved As-Projective-As-Possible (APAP) algorithm was designed to extract and stitch keyframes from videos. The graph convolutional neural network (GCN) was used to segment the stitched image. The GCN’s m-IOU is 24.02% higher than Fully convolutional networks (FCN), proving that GCN has great potential of application in image segmentation and underwater image processing. The result shows that the improved APAP algorithm and GCN can adapt to complex underwater environments and perform well in different study areas.

关键词： underwater cracks remote operated vehicle image stitching image segmentation graph convolutional neural network

HTML PDF 收藏

基于回归预测集成学习的交互式图像分割 Article

Jin ZHANG, Zhao-hui TANG, Wei-hua GUI, Qing CHEN, Jin-ping LIU

《信息与电子工程前沿（英文）》 2017年第18卷第7期页码 1002-1020 doi: 10.1631/FITEE.1601401

摘要：对于复杂场景下的自然图像，全自动图像分割方法难以获得与真实情况吻合的结果，人们常常采用交互式分割手段实现精确分割。然而，当前及背景中存在颜色相似的区域时，传统半监督图像分割方法只能通过大量增加手工标记获得精确分割结果。为此，本文提出一种结合半监督学习的基于回归预测的集成学习交互式图像分割方法。通过集成两个互补的样条回归函数，将图像分割视为一个非线性预测问题。首先，基于已标记样本训练出两个在属性上互补的多元自适应回归样条学习器（multivariate adaptive regression splines, MARS）和薄板样条回归学习器（thin plate spline regression, TPSR）；接着，提出一种基于聚类假设和半监督学习的回归器增强算法，该算法从未标记样本中抽选部分样本辅助训练MARS和TPSR；然后，引入支持向量回归方法（support vector regression, SVR）集成MARS和TPSR的预测结果；最后，对SVR集成结果进行GraphCut图像分割。在标准数据库BSDS500和Pascal VOC上进行大量实验，验证了所提算法的有效性。大量对比实验证实，所提算法在交互式自然图像分割上的表现与当前最先进算法相当。

关键词：交互式图像分割；多元自适应回归样条；集成学习；薄板样条回归；半监督学习；支持向量回归

HTML PDF 收藏

基于自适应置信度校准的交互式医疗图像分割框架

沈楚云,李文浩,徐琪森,胡斌,金博,蔡海滨,朱凤平,李郁欣,王祥丰

《信息与电子工程前沿（英文）》 2023年第24卷第9期页码 1332-1348 doi: 10.1631/FITEE.2200299

摘要：基于人机交互的医疗图像分割方法是一种新的范式，其通过引入专家交互信息来指导算法完成图像分割任务。然而，现有医疗图像分割模型往往容易产生“交互误解”，即无法合理权衡短期和长期交互信息的重要性。为更好地利用不同时间尺度上的交互信息，本文提出一种基于自适应置信度校准的交互式医疗图像分割框架MECCA，其结合了基于分割决策的置信度学习技术和多智能体强化学习技术，并通过预测分割决策与短期交互信息的对齐水平来学习一个新颖的置信度网络。随后，提出一种基于置信度的奖励塑造机制，在策略梯度计算中引入置信度，从而直接纠正模型产生的交互误解。MECCA还通过标签生成和交互指导来降低交互强度和难度，从而实现用户友好交互。实验结果表明，MECCA在不同分割任务中可以显著提高短期和长期交互信息的利用效率，且仅需较少的标注样本。演示视频可通过https://bit.ly/mecca-demo-video访问。

关键词：医疗图像分割交互式分割多智能体强化学习置信度学习半监督学习

HTML 收藏

Deep learning in digital pathology image analysis: a survey

Shujian Deng, Xin Zhang, Wen Yan, Eric I-Chao Chang, Yubo Fan, Maode Lai, Yan Xu

《医学前沿（英文）》 2020年第14卷第4期页码 470-487 doi: 10.1007/s11684-020-0782-9

摘要： deep learning (DL) has achieved state-of-the-art performance in many digital pathology analysis tasks. Traditional methods usually require hand-crafted domain-specific features, and DL methods can learn representations without manually designed features. In terms of feature extraction, DL approaches are less labor intensive compared with conventional machine learning methods. In this paper, we comprehensively summarize recent DL-based image analysis studies in histopathology, including different tasks (e.g., classification, semantic segmentation, detection, and instance segmentation) and various applications (e.g., stain normalization, cell/gland/region structure analysis). DL methods can provide consistent and accurate outcomes. DL is a promising tool to assist pathologists in clinical diagnosis.

关键词： pathology deep learning segmentation detection classification

HTML PDF 收藏

基于边界分析的森林冠层半球图像中心点定位与分割 Article

Jia-yin SONG,Wen-long SONG,Jian-ping HUANG,Liang-kuan ZHU

《信息与电子工程前沿（英文）》 2016年第17卷第8期页码 741-749 doi: 10.1631/FITEE.1601169

摘要：分析森林半球图像是测定森林冠层结构参数的重要方法之一。本文主要研究半球图像中圆形区域的分割方法，这是分析半球图像的基础。通过直方图、矩形度和傅里叶描述子进行森林半球图像边界的分析，得到边界特性如下：(1)边缘模型包含三种，分别是台阶、斜坡和屋顶边缘模型；(2)边界点离散；(3)边缘存在线性分布离散点。在此基础上我们提出了森林半球图像圆形区域的分割方法，拟合圆形边界线，同时用最小二乘法计算圆心点坐标及半径。该方法与获取图像的硬件设备参数无关，因此为引入参数自动调整的高性能设备获取森林半球图像奠定了基础。

关键词：鱼眼镜头；最小二乘法；图像分割；生态学图像处理；半球图像

HTML PDF 收藏

过渡区提取方法综述

刘锁兰,杨静宇

《中国工程科学》 2007年第9卷第9期页码 89-96

摘要：

图像分割是图像理论发展的瓶颈，过渡区是指图像中介于目标和背景之间的特殊区域，借助于过渡区的确定进行图像分割。主要介绍了过渡区提取的两大类方法：基于梯度的方法和基于非梯度的方法，并对提取效果以及存在的问题做了简要分析。

关键词：过渡区提取图像分割梯度法非梯度法

HTML PDF 收藏

基于双层多目标分割的超高速撞击航天器损伤红外检测算法 Research Article

杨晓1,殷春1,Sara DADRAS2,雷光钰1,谭旭彤1,邱根1

《信息与电子工程前沿（英文）》 2022年第23卷第4期页码 571-586 doi: 10.1631/FITEE.2000695

摘要：针对超高速撞击引起的航天器损伤检测，提出一种先进的基于红外成像检测的航天器缺陷提取算法。采用高速混合模型对红外视频流采样数据中的温度变化特征进行分类，并重构图像，得到反映缺陷特征的红外重构图像。设计的分割目标函数用于保证图像分割结果对噪声去除和细节保留的有效性，同时考虑到红外重构图像的复杂性，即所需权衡不同。因此，引入多目标优化算法以实现细节保留和噪声去除之间的平衡，并采用基于分解的多目标进化算法（MOEA/D）进行优化，以保证损伤分割的准确性。实验结果验证了所提算法的有效性。

关键词：超高速撞击损伤；缺陷检测；高斯混合模型；图像分割

HTML PDF 收藏

Turbidity-adaptive underwater image enhancement method using image fusion

《机械工程前沿（英文）》 2022年第17卷第3期 doi: 10.1007/s11465-021-0669-8

摘要： Clear, correct imaging is a prerequisite for underwater operations. In real freshwater environment including rivers and lakes, the water bodies are usually turbid and dynamic, which brings extra troubles to quality of imaging due to color deviation and suspended particulate. Most of the existing underwater imaging methods focus on relatively clear underwater environment, it is uncertain that if those methods can work well in turbid and dynamic underwater environments. In this paper, we propose a turbidity-adaptive underwater image enhancement method. To deal with attenuation and scattering of varying degree, the turbidity is detected by the histogram of images. Based on the detection result, different image enhancement strategies are designed to deal with the problem of color deviation and blurring. The proposed method is verified by an underwater image dataset captured in real underwater environment. The result is evaluated by image metrics including structure similarity index measure, underwater color image quality evaluation metric, and speeded-up robust features. Test results exhibit that the method can correct the color deviation and improve the quality of underwater images.

关键词： turbidity underwater image enhancement image fusion underwater robots visibility

HTML PDF 收藏

基于核稀疏表示的磁共振图像分析及其在脑肿瘤自动分割中的应用 None

Ji-jun TONG, Peng ZHANG, Yu-xiang WENG, Dan-hua ZHU

《信息与电子工程前沿（英文）》 2018年第19卷第4期页码 471-480 doi: 10.1631/FITEE.1620342

摘要：脑肿瘤分割在疾病辅助诊断、治疗方案规划以及手术导航中扮演重要角色。对脑肿瘤精确分割可以帮助临床医生获取肿瘤位置、尺寸和形状信息。提出一种基于核稀疏编码的全自动脑肿瘤分割方法，并在3D多模态磁共振成像图（magnetic resonance imaging, MRI）上验证。首先对MRI图像进行预处理以减少噪声，然后通过核字典学习提取非线性特征，用来构建坏死组织、水肿组织、非增强肿瘤组织、增强肿瘤组织和健康组织5个适应性字典。对从原始MRI图像上肿瘤像素点周边m×m×m的小区域提取的特征向量进行稀疏编码，并通过一种基于字典学习的核聚类方法对像素点进行编码。最后通过形态滤波填充在多个相连部分间的区域，提高分割质量。为评估分割表现，分割结果被上传到在线评估系统中，该评估系统使用dice系数、阳性预测值（positive predictive value, PPV）、灵敏度和kappa值作为评估指标。结果表明，该方法在完整肿瘤区域分割上具有良好表现（dice: 0.83; PPV: 0.84; sensitivity: 0.82），而在肿瘤核心区域（dice: 0.69; PPV: 0.76; sensitivity: 0.80）和增强肿瘤区域（dice: 0.58; PPV: 0.60; sensitivity: 0.65）上表现稍差。相较于脑肿瘤分割（BRATS）挑战中其他团队采用的方法，该方法具有竞争力。该方法在健康组织和病理组织区分上具有一定潜力。

关键词：脑肿瘤分割；核方法；稀疏编码；字典学习

HTML PDF 收藏

Gradient-based compressive image fusion

Yang CHEN,Zheng QIN

《信息与电子工程前沿（英文）》 2015年第16卷第3期页码 227-237 doi: 10.1631/FITEE.1400217

摘要： We present a novel image fusion scheme based on gradient and scrambled block Hadamard ensemble (SBHE) sampling for compressive sensing imaging. First, source images are compressed by compressive sensing, to facilitate the transmission of the sensor. In the fusion phase, the image gradient is calculated to reflect the abundance of its contour information. By compositing the gradient of each image, gradient-based weights are obtained, with which compressive sensing coefficients are achieved. Finally, inverse transformation is applied to the coefficients derived from fusion, and the fused image is obtained. Information entropy (IE), Xydeas’s and Piella’s metrics are applied as non-reference objective metrics to evaluate the fusion quality in line with different fusion schemes. In addition, different image fusion application scenarios are applied to explore the scenario adaptability of the proposed scheme. Simulation results demonstrate that the gradient-based scheme has the best performance, in terms of both subjective judgment and objective metrics. Furthermore, the gradient-based fusion scheme proposed in this paper can be applied in different fusion scenarios.

关键词： Compressive sensing (CS) Image fusion Gradient-based image fusion CS-based image fusion

HTML PDF 收藏

Edge detection of steel plates at high temperature using image measurement

Qiong Zhou, Qi An

《机械工程前沿（英文）》 2009年第4卷第1期页码 77-82 doi: 10.1007/s11465-009-0013-1

摘要： An edge detection method for the measurement of steel plate’s thermal expansion is proposed in this paper, where the shrinkage of a steel plate is measured when temperature drops. First, images are picked up by an imaging system; a method of regional edge detection based on grayscales’ sudden change is then applied to detect the edges of the steel plate; finally, pixel coordinates of the edge position are transformed to physical coordinates through calibration parameters. The experiment shows that the real-time, high precision, and non-contact measurement of the steel plate’s edge position under high temperature can be realized using the imaging measurement method established in this paper.

关键词： thermal expansion image measurement edge detection image calibration

HTML PDF 收藏

Fast detection algorithm for cracks on tunnel linings based on deep semantic segmentation

《结构与土木工程前沿（英文）》 2023年第17卷第5期页码 732-744 doi: 10.1007/s11709-023-0965-y

摘要： An algorithm based on deep semantic segmentation called LC-DeepLab is proposed for detecting the trends and geometries of cracks on tunnel linings at the pixel level. The proposed method addresses the low accuracy of tunnel crack segmentation and the slow detection speed of conventional models in complex backgrounds. The novel algorithm is based on the DeepLabv3+ network framework. A lighter backbone network was used for feature extraction. Next, an efficient shallow feature fusion module that extracts crack features across pixels is designed to improve the edges of crack segmentation. Finally, an efficient attention module that significantly improves the anti-interference ability of the model in complex backgrounds is validated. Four classic semantic segmentation algorithms (fully convolutional network, pyramid scene parsing network, U-Net, and DeepLabv3+) are selected for comparative analysis to verify the effectiveness of the proposed algorithm. The experimental results show that LC-DeepLab can accurately segment and highlight cracks from tunnel linings in complex backgrounds, and the accuracy (mean intersection over union) is 78.26%. The LC-DeepLab can achieve a real-time segmentation of 416 × 416 × 3 defect images with 46.98 f/s and 21.85 Mb parameters.

关键词： tunnel engineering crack segmentation fast detection DeepLabv3+ feature fusion attention mechanism

HTML PDF 收藏

Deformable image registration with geometric changes

Yu LIU,Bo ZHU

《信息与电子工程前沿（英文）》 2015年第16卷第10期页码 829-837 doi: 10.1631/FITEE.1500045

摘要： Geometric changes present a number of difficulties in deformable image registration. In this paper, we propose aglobal deformation framework to model geometric changes whilst promoting a smooth transformation between source and target images. To achieve this, we have developed an innovative model which significantly reduces the side effects of geometric changes in image registration, and thus improves the registration accuracy. Our key contribution is the introduction of a sparsity-inducing norm, which is typically L1 norm regularization targeting regions where geometric changes occur. This preserves the smoothness of global transformation by eliminating local transformation under different conditions. Numerical solutions are discussed and analyzed to guarantee the stability and fast convergence of our algorithm. To demonstrate the effectiveness and utility of this method, we evaluate it on both synthetic data and real data from traumatic brain injury (TBI). We show that the transformation estimated from our model is able to reconstruct the target image with lower instances of error than a standard elastic registration model.

关键词： Geometric changes Image registration Sparsity Traumatic brain injury (TBI)

HTML PDF 收藏

Multi-color space threshold segmentation and self-learning k-NN algorithm for surge test EUT status

Jian HUANG,Gui-xiong LIU

《机械工程前沿（英文）》 2016年第11卷第3期页码 311-315 doi: 10.1007/s11465-016-0376-z

摘要：

The identification of targets varies in different surge tests. A multi-color space threshold segmentation and self-learning k-nearest neighbor algorithm (k-NN) for equipment under test status identification was proposed after using feature matching to identify equipment status had to train new patterns every time before testing. First, color space (L*a*b*, hue saturation lightness (HSL), hue saturation value (HSV)) to segment was selected according to the high luminance points ratio and white luminance points ratio of the image. Second, the unknown class sample S_r was classified by the k-NN algorithm with training set T_z according to the feature vector, which was formed from number of pixels, eccentricity ratio, compactness ratio, and Euler’s numbers. Last, while the classification confidence coefficient equaled k, made S_r as one sample of pre-training set T_z′. The training set T_z increased to T_z₊₁ by T_z′ if T_z′ was saturated. In nine series of illuminant, indicator light, screen, and disturbances samples (a total of 21600 frames), the algorithm had a 98.65% identification accuracy, also selected five groups of samples to enlarge the training set from T₀ to T₅ by itself.