结合SAM视觉分割模型与随机森林机器学习的无人机影像盐沼植被“精灵圈”提取

“精灵圈”是海岸带盐沼植被生态系统中的一种“空间自组织”结构,对盐沼湿地的生产力、稳定性和恢复力有重要影响。无人机影像是实现“精灵圈”空间位置高精度识别及解译其时空演化趋势与规律的重要数据源,但“精灵圈”像素与背景像素在色彩信息和外形特征上差异较小,如何从二维影像中智能精准地识别“精灵圈”像素并对识别的单个像素形成个体“精灵圈”是目前的技术难点。本文提出了一种结合分割万物模型(Segment Anything Model,SAM)视觉分割模型与随机森林机器学习的无人机影像“精灵圈”分割及分类方法,实现了单个“精灵圈”的识别和提取。首先,通过构建索伦森-骰子系数(S?rensen-Dice coefficient,Dice)和交并比(Intersection over Union,IOU)评价指标,从SAM中筛选预训练模型并对其参数进行优化,实现全自动影像分割,得到无属性信息的分割掩码/分割类;然后,利用红、绿、蓝(RGB)三通道信息及空间二维坐标将分割掩码与原图像进行信息匹配,构造分割掩码的特征指标,并根据袋外数据(Out of Bag,OOB)误差减小及特征分布规律对特征进行分析和筛选;最后,利用筛选的特征对随机森林模型进行训练,实现“精灵圈”植被、普通植被和光滩的自动识别与分类。实验结果表明:本文方法“精灵圈”平均正确提取率96.1%,平均错误提取率为9.5%,为精准刻画“精灵圈”时空格局及海岸带无人机遥感图像处理提供了方法和技术支撑。

一种街景图像中建筑物高度估算方法

建筑物高度信息是城市三维建模的基础数据,但已有的建筑物高度估算研究多采用LiDAR和SAR等遥感影像。随着计算机和互联网的快速发展,街景数据因采集容易和成本低等特点成为了一种新兴的建筑物高度估算数据源。文章提出一种街景图像中建筑物高度估算方法,首先利用segment anything model实现图像中建筑物像素高度提取;然后利用图像元数据和电子地图数据获取建筑物与相机之间的距离、图像焦距,根据街景图像与建筑物实体的几何关系改进针孔相机模型,构建建筑物高度估算方法;最后选取北京、柏林的Mapillary街景图像开展实验验证。结果表明,与改进前相比,改进后针孔相机模型明显提升了高度估算准确度,RMSE降低了11.31 m,R^(2)提高了0.4,具备实用价值。

基于SAM&ImageJ图像处理的堆石混凝土坝层面露石率研究

堆石混凝土坝层面的外露块石为上下层提供了重要的啮合作用,其投影面积比例是科学评价层间抗剪性能的重要指标。采用国际最新Meta AI模型segment anything model(SAM)对层面外露堆石进行自动图像分割,并基于ImageJ软件对SAM识别后的图片进行再加工与图像计算,利用平滑、差分算法、中值滤波等方法精准标定外露堆石,二值化后计算得到层面露石率。结果表明:SAM图像预分割可识别约90%的外露堆石,经过ImageJ二次图像处理后可有效提高小粒径堆石的识别精度,对比手动标注结果误差在±3%以内。以贵州省两座水库的工程应用为例,对浇筑仓面进行分区预处理,结果发现靠近上游、中部、下游不同区域的露石率差别较大,计算得到的层面露石率以10%~30%居多,其中堆石入仓运输通道区域的露石率较低。研究内容与结论可为堆石混凝土结构层间界面抗剪力学性能和大坝蓄水安全稳定的研究提供参考与借鉴。

Analysis and comparison of retinal vascular parameters under different glucose metabolic status based on deep learning

AIM:To develop a deep learning-based model for automatic retinal vascular segmentation,analyzing and comparing parameters under diverse glucose metabolic status(normal,prediabetes,diabetes)and to assess the potential of artificial intelligence(AI)in image segmentation and retinal vascular parameters for predicting prediabetes and diabetes.METHODS:Retinal fundus photos from 200 normal individuals,200 prediabetic patients,and 200 diabetic patients(600 eyes in total)were used.The U-Net network served as the foundational architecture for retinal arteryvein segmentation.An automatic segmentation and evaluation system for retinal vascular parameters was trained,encompassing 26 parameters.RESULTS:Significant differences were found in retinal vascular parameters across normal,prediabetes,and diabetes groups,including artery diameter(P=0.008),fractal dimension(P=0.000),vein curvature(P=0.003),C-zone artery branching vessel count(P=0.049),C-zone vein branching vessel count(P=0.041),artery branching angle(P=0.005),vein branching angle(P=0.001),artery angle asymmetry degree(P=0.003),vessel length density(P=0.000),and vessel area density(P=0.000),totaling 10 parameters.CONCLUSION:The deep learning-based model facilitates retinal vascular parameter identification and quantification,revealing significant differences.These parameters exhibit potential as biomarkers for prediabetes and diabetes.

Material-SAM:Adapting SAM for Material XCT

X-ray Computed Tomography(XCT)enables non-destructive acquisition of the internal structure of materials,and image segmentation plays a crucial role in analyzing material XCT images.This paper proposes an image segmentation method based on the Segment Anything model(SAM).We constructed a dataset of carbide in nickel-based single crystal superalloys XCT images and preprocessed the images using median filtering,histogram equalization,and gamma correction.Subsequently,SAM was fine-tuned to adapt to the task of material XCT image segmentation,resulting in Material-SAM.We compared the performance of threshold segmentation,SAM,U-Net model,and Material-SAM.Our method achieved 88.45%Class Pixel Accuracy(CPA)and 88.77%Dice Similarity Coefficient(DSC)on the test set,outperforming SAM by 5.25%and 8.81%,respectively,and achieving the highest evaluation.Material-SAM demonstrated lower input requirements compared to SAM,as it only required three reference points for completing the segmentation task,which is one-fifth of the requirement of SAM.Material-SAM exhibited promising results,highlighting its potential as a novel method for material XCT image segmentation.

Theoretical analysis of a new segmented anchoring style in weakly cemented soft surrounding rock

According to the tensile failure of rock bolt in weakly cemented soft rock, this paper presents a new segmented anchoring style in order to weaken the cumulative effect of anchoring force associated with the large deformation. Firstly, a segmented mechanical model was established in which free and anchoring section of rock bolt were respectively arranged in different deformation zones. Then, stress and displacement in elastic non-anchoring zone, elastic anchoring zone, elastic sticking zone, softening sticking zone and broken zone were derived respectively based on neural theory and tri-linear strain softening constitutive model of soft rock. Results show that the anchoring effect can be characterized by a supporting parameter b. With its increase, the peak value of tangential stress gradually moves to the roadway wall, and the radial stress significantly increases, which means the decrease of equivalent plastic zone and improvement of confining effect provided by anchorage body. When b increases to 0.72, the equivalent plastic zone disappears, and stresses tend to be the elastic solutions. In addition, the anchoring effect on the displacement of surrounding rock can be quantified by a normalization factor δ.

APPROXIMATION TECHNIQUES FOR APPLICATION OF GENETIC ALGORITHMS TO STRUCTURAL OPTIMIZATION

Although the genetic algorithm (GA) has very powerful robustness and fitness, it needs a large size of population and a large number of iterations to reach the optimum result. Especially when GA is used in complex structural optimization problems, if the structural reanalysis technique is not adopted, the more the number of finite element analysis (FEA) is, the more the consuming time is. In the conventional structural optimization the number of FEA can be reduced by the structural reanalysis technique based on the approximation techniques and sensitivity analysis. With these techniques, this paper provides a new approximation model-segment approximation model, adopted for the GA application. This segment approximation model can decrease the number of FEA and increase the convergence rate of GA. So it can apparently decrease the computation time of GA. Two examples demonstrate the availability of the new segment approximation model.

Stem taper equations for diameter and volume predictions of Abies cilicica Carr.in the Taurus Mountains,Turkey

Taper equations are an important tool in estimating stem volumes at a multi-product level for sustainable forest management.Nine taper equations are tested from three categories to estimate diameter at specified point on the stem,height at specified diameter,volumes of any desired portion of the stem,and whole tree volume of Taurus fir in Taurus Mountains of Turkey.To account for autocorrelation and multicollinearity present among multiple stem data observations collected from the same tree.proper statistical approaches were used in model fitting.Comparisons are made to determine which equation provides the best overall fit to all data based on four goodness-of-fit statistics,Coefficient of determination(R2);Root mean square error(RMSE),Akaike’s Information Criterion(AIC),and Bayesian information criterion(BIC).Results indicated that all taper equations tested could be used to accurately estimate section diameter at given height and stem volume.Clark et al.’s taper equation provided better results than the others for Taurus fir when an additional stem diameter observation at 5.30 m was available.Segmented and variable-form taper models consistently provided better results than the simple taper models except for Max and Burkhart’s model.Fang et al.’s and Kozak’s taper models showed equally good performance to describe stem taper and to predict tree stem volume.Therefore,these taper equations are able to be used to estimate diameter and volume for Taurus fir trees,if an upper stem diameter measurement was not available.

Compatible taper and stem volume equations for Larix kaempferi(Japanese larch) species of South Korea

In this study, compatible taper and stem volume equations were developed for Larix kaempferi species of South Korea. The dataset was split into two groups： 80% of the data were used in model fitting and the remaining 2o% were used for validation. The compatible MB76 equations were used to predict the diameter outside bark to a specific height, the height to a specific diameter and the stem volume of the species. The result of the stem volume analysis was compared with the existing stem volume model of Larix kaempferi species of South Korea which was developed by the Korea Forest Research Institute and with a simple volume model that was developed with fitting dataset in this study. The compatible model provided accurate prediction of the total stem volume when compared to the existing stem volume model and with a simple volume model. It is concluded that the compatible taper and stem volume equations are more convenient to use and therefore it is recommended to be applied in the Larix kaempferi species of South Korea.

Development of ecoregion-based merchantable volume systems for Pinus brutia Ten. and Pinus nigra Arnold. in southern Turkey

Estimating individual tree volume is one of the essential building blocks in forest growth and yield models. Ecologically based taper equations provide accurate vol- ume predictions and allow classification by mer- chantable sizes, assisting in sustainable forest management. In the present study, ecoregion-based compatible volume systems for brutian pine and black pine in the three ecoregions of southern Turkey were developed. Several well-known taper functions were evaluated. A second- order continuous-time autoregressive error structure was used to correct the inherent autocorrelation in the hierar- chical data, allowing the model to be applied to irregularly spaced and unbalanced data. The compatible segmented model of Fang et al. （For Sci 46：1-12, 2000） best described the experimental data. It is therefore recommended for estimating diameter at a specific height, height to a specific diameter, merchantable volume, and total volume for the three ecoregions and two species analyzed. The nonlinearextra sum of squares method indicated differences in ecoregion and tree-specific taper functions. A different taper function should therefore be used for each pine spe- cies and ecoregion in southern Turkey. Using ecoregion- specific taper equations allows making more robust esti- mations and, therefore, will enhance the accuracy of diameter at different heights and volume predictions.

FTIR STUDIES ON THE MODEL POLYURETHANE HARD SEGMENTS BASED ON A NEW WATERBORNE CHAIN EXTENDER DIMETHYLOL BUTANOIC ACID (DMBA)

Three model polyurethane hard segments based on dimethylol butanoic acid (DMBA) and 1,6-hexane diisocyanate (HDI), toluene diisocyanate (TDI) and 4,4'-diphenylmethane diisocyanate (MDI) were prepared by the solution method. Fourier Infrared (FTIR) spectroscopy was employed to study the H-bonds in these model polyurethanes. The model polyurethane hard segment prepared from HDI and 1,4-butanodiol (BDO) was used for comparison. It was found that the incorporation of the pendent carboxyl through DMBA into the model hard segments weakens the original NH…O = C H-bond but gives more H-bond patterns based on the two H-bond donors, urethane NH and carboxylic OH. The carboxylic dimer is one of the main H-bond types and is stronger than another main H-bond type NH…O=C. In addition, the H-bond in aromatic model hard segments is stronger than that of aliphatic hard segments. The appearance of the free C=O and the fact that almost all N—H is H-bonded suggest that there possibly exist either the third H-bond acceptor or the H-bond formed by one acceptor with two donors.

Stem taper equations for poplars growing on farmland in Sweden

We developed a simple polynomial taper equation for poplars growing on former farmland in Sweden and also evaluated the performance of some well-known taper equations. In Sweden there is an increasing interest in the use of poplar. Effective management of poplar plantations for high yield production would be facilitated by taper equations providing better predictions of stem volume than currently available equations. In the study a polynomial stem taper equation with five parameters was established for individual poplar trees growing on former farmland. The outputs of the polynomial taper equation were compared with five published equations. Data for fitting the equations were collected from 69 poplar trees growing at 37 stands in central and southern Sweden （lat. 55–60° N）. The mean age of the stands was 21 years （range 14–43）, the mean density 984 stems·ha？1 （198–3,493）, and the mean diameter at breast height （outside bark） 25 cm （range 12–40）. To verify the tested equations, performance of accuracy and precision diameter predictions at seven points along the stem was closely analyzed. Statistics used for evaluation of the equations indicated that the variable exponent taper equation presented by Kozak （1988） performed best and can be recommended. The stem taper equation by Kozak （1988） recommended in the study is likely to be beneficial for optimising the efficiency and profitability of poplar plantation management. The constructed polynomial equation and the segmented equation presented by Max ＆ Burkhart （1976） were second and third ranked. Due to the statistical complexity of Kozak’s equation, the constructed polynomial equation is alternatively recommended when a simple model is requested and larger bias is accepted.

Estimation of Springing Response for 550 000 DWT Ore Carrier

The desire to benefit from economy of scale is one of the major driving forces behind the continuous growth in ship sizes. However, models of new large ships need to be thoroughly investigated to determine the carrier's response in waves. In this work, experimental and numerical assessments of the motion and load response of a 550,000 DWT ore carrier are performed using prototype ships with softer stiffness, and towing tank tests are conducted using a segmented model with two schemes of softer stiffness. Numerical analyses are performed employing both rigid body and linear hydroelasticity theories using an in-house program and a comparison is then made between experimental and numerical results to establish the influence of stiffness on the ore carrier's springing response. Results show that softer stiffness models can be used when studying the springing response of ships in waves.

Co-seismic Coulomb stress changes on the northern Tanlu fault zone caused by the Tohoku-Oki M_(W)9.0 earthquake

Based on the spherical earth dislocation theory and a fault slip model of the Tohoku-Oki M_(W)9.0 earthquake,the co-seismic Coulomb failure stress changes(ΔCFS)on the northern Tanlu fault zone at depths of 0–40 km are calculated.By comparing two sets of results from the spherical earth dislocation theory and the semi-infinite space one,the effect of earth curvature on the calculation results is analyzed quantitatively.First,we systematically summarize previous researches related to the northern Tanlu fault zone,divide the fault zone as detailed as possible,give the geometric parameters of each segment,and establish a segmented structural model of the northern Tanlu fault zone.Second,we calculate the Coulomb stress changes on the northern Tanlu fault zone by using the spherical earth dislocation theory.The result shows the Coulomb stress changes are no more than 0.003 MPa,which proves the great earthquake did not significantly change the stress state of the fault zone.Finally,we quantitatively analyze the disparities between the results of semi-infinite space dislocation theory and the spherical earth one.The average disparity between them is about 7.7%on the northern Tanlu fault zone and is 16.8%on the Fangzheng graben,the maximum disparity on this graben reaches up to 25.5%.It indicates that the effect of earth curvature can not be ignored.So it’s necessary to use the spherical earth dislocation theory instead of the semi-infinite space one to study the Coulomb stress change in the far field.

Wind tunnel study of aerodynamic wind loading on middle pylon of Taizhou Bridge

Segment sectional model tests are carried out to investigate the wind loading on middle pylon of Taizhou Bridge, which has complicated three-dimensional flow due to its feature of double columns. Through the force measuring tests, aerodynamic force coefficients of every segment of the pylon columns have been obtained. It is found that the tested aerodynamic force coefficients are much smaller than those given by codes. The interference effects of aerodynamic force coefficients between columns of pylon are discussed. The results show that the interference effect is the most evident when the yaw angle is about 30 ° from transverse direction. This kind of interference effect can be described as diminutions in transverse aerodynamic force coefficients and magnifications in longitudinal aerodynamic force coefficients of downstream columns.

Implicit Shape Reconstruction of Unorganized Points Using PDE-Based Deformable 3D Manifolds

In this work we consider the problem of shape reconstruction from an unorganized data set which has many important applications in medical imaging, scientific computing, reverse engineering and geometric modelling. The reconstructed surface is obtained by continuously deforming an initial surface following the Partial Differential Equation (PDE)-based diffusion model derived by a minimal volume-like variational formulation. The evolution is driven both by the distance from the data set and by the curvature analytically computed by it. The distance function is computed by implicit local interpolants defined in terms of radial basis functions. Space discretization of the PDE model is obtained by finite co-volume schemes and semi-implicit approach is used in time/scale. The use of a level set method for the numerical computation of the surface reconstruction allows us to handle complex geometry and even changing topology,without the need of user-interaction. Numerical examples demonstrate the ability of the proposed method to produce high quality reconstructions. Moreover, we show the effectiveness of the new approach to solve hole filling problems and Boolean operations between different data sets.

Preface

Micromechanics aims mainly at establishing the quantitative relation between the macroscopic mechanical behavior and the microstructure of heterogeneous materials.

A nonlinear hydroelastic method considering wave memory effect for ship load responses in irregular waves

Since the amplitude and frequency of irregular waves change with time,great difficulties are brought for solving ship load responses in random waves.To take the effect of various frequencies of irregular waves into consideration in load responses of hull,the wave memory effect is necessary.A semi-analytical method is introduced for the time-domain retardation functions,and then a nonlinear hydroelastic method considering memory effect for ships in irregular waves is proposed.Segmented self-propelling model experiments of a container ship were carried out in a towing tank,a ship motion measuring device for self-propelling model test was designed.Whipping responses of the ship in regular and irregular waves are analyzed.Finally,the calculation results are compared with those measured by segmented model experiments,and the result indicates that the memory effect has little effect on load responses of ship in regular waves,but pronounced effect on results in irregular waves.Moreover,the presented method is reasonable for the prediction of ship load responses in irregular waves.

The use of the greater trochanter marker in the thigh segment model:Implications for hip and knee frontal and transverse plane motion

Background:The greater trochanter marker is commonly used in 3-dimensional(3D) models;however,its influence on hip and knee kinematics during gait is unclear.Understanding the influence of the greater trochanter marker is important when quantifying frontal and transverse plane hip and knee kinematics,parameters which are particularly relevant to investigate in individuals with conditions such as patellofemoral pain,knee osteoarthritis,anterior cruciate ligament(ACL) injury,and hip pain.The aim of this study was to evaluate the effect of including the greater trochanter in the construction of the thigh segment on hip and knee kinematics during gait.Methods:3D kinematics were collected in 19 healthy subjects during walking using a surface marker system.Hip and knee angles were compared across two thigh segment definitions(with and without greater trochanter) at two time points during stance:peak knee flexion(PKF) and minimum knee flexion(Min KF).Results:Hip and knee angles differed in magnitude and direction in the transverse plane at both time points.In the thigh model with the greater trochanter the hip was more externally rotated than in the thigh model without the greater trochanter(PKF:-9.34°± 5.21° vs.1.40°± 5.22°,Min KF:-5.68°± 4.24° vs.5.01°± 4.86°;p < 0.001).In the thigh model with the greater trochanter,the knee angle was more internally rotated compared to the knee angle calculated using the thigh definition without the greater trochanter(PKF:14.67°± 6.78° vs.4.33°± 4.18°,Min KF:10.54°± 6.71° vs.-0.01°± 2.69°;p < 0.001).Small but significant differences were detected in the sagittal and frontal plane angles at both time points(p < 0.001).Conclusion:Hip and knee kinematics differed across different segment definitions including or excluding the greater trochanter marker,especially in the transverse plane.Therefore when considering whether to include the greater trochanter in the thigh segment model when using a surface markers to calculate 3D kinematics for movement assessment,it is important to have a clear understanding of the effect of different marker sets and segment models in use.

Segment Anything Is Not Always Perfect: An Investigation of SAM on Different Real-world Applications

Recently,Meta AI Research approaches a general,promptable segment anything model(SAM)pre-trained on an unprecedentedly large segmentation dataset(SA-1B).Without a doubt,the emergence of SAM will yield significant benefits for a wide array of practical image segmentation applications.In this study,we conduct a series of intriguing investigations into the performance of SAM across various applications,particularly in the fields of natural images,agriculture,manufacturing,remote sensing and healthcare.We analyze and discuss the benefits and limitations of SAM,while also presenting an outlook on its future development in segmentation tasks.By doing so,we aim to give a comprehensive understanding of SAM's practical applications.This work is expected to provide insights that facilitate future research activities toward generic segmentation.Source code is publicly available at https://github.com/LiuTingWed/SAM-Not-Perfect.