Infrared and Visible Image Fusion Based on Res2Net-Transformer Automatic Encoding and Decoding

A novel image fusion network framework with an autonomous encoder and decoder is suggested to increase thevisual impression of fused images by improving the quality of infrared and visible light picture fusion. The networkcomprises an encoder module, fusion layer, decoder module, and edge improvementmodule. The encoder moduleutilizes an enhanced Inception module for shallow feature extraction, then combines Res2Net and Transformerto achieve deep-level co-extraction of local and global features from the original picture. An edge enhancementmodule (EEM) is created to extract significant edge features. A modal maximum difference fusion strategy isintroduced to enhance the adaptive representation of information in various regions of the source image, therebyenhancing the contrast of the fused image. The encoder and the EEM module extract features, which are thencombined in the fusion layer to create a fused picture using the decoder. Three datasets were chosen to test thealgorithmproposed in this paper. The results of the experiments demonstrate that the network effectively preservesbackground and detail information in both infrared and visible images, yielding superior outcomes in subjectiveand objective evaluations.

A color image encryption scheme based on a 2D coupled chaotic system and diagonal scrambling algorithm

A novel color image encryption scheme is developed to enhance the security of encryption without increasing the complexity. Firstly, the plain color image is decomposed into three grayscale plain images, which are converted into the frequency domain coefficient matrices(FDCM) with discrete cosine transform(DCT) operation. After that, a twodimensional(2D) coupled chaotic system is developed and used to generate one group of embedded matrices and another group of encryption matrices, respectively. The embedded matrices are integrated with the FDCM to fulfill the frequency domain encryption, and then the inverse DCT processing is implemented to recover the spatial domain signal. Eventually,under the function of the encryption matrices and the proposed diagonal scrambling algorithm, the final color ciphertext is obtained. The experimental results show that the proposed method can not only ensure efficient encryption but also satisfy various sizes of image encryption. Besides, it has better performance than other similar techniques in statistical feature analysis, such as key space, key sensitivity, anti-differential attack, information entropy, noise attack, etc.

Image Dehazing with Hybrid λ2-λ0 Penalty Mode

Due to the presence of turbid media, such as microdust and water vapor in the environment, outdoor pictures taken under hazy weather circumstances are typically degraded. To enhance the quality of such images, this work proposes a new hybrid λ2-λ0 penalty model for image dehazing. This model performs a weighted fusion of two distinct transmission maps, generated by imposing λ2 and λ0 norm penalties on the approximate regression coefficients of the transmission map. This approach effectively balances the sparsity and smoothness associated with the λ0 and λ2 norms, thereby optimizing the transmittance map. Specifically, when the λ2 norm is penalized in the model, an updated guided image is obtained after implementing λ0 penalty. The resulting optimization problem is effectively solved using the least square method and the alternating direction algorithm. The dehazing framework combines the advantages of λ2 and λ0 norms, enhancing sparse and smoothness, resulting in higher quality images with clearer details and preserved edges.

MRI T_2 star mapping、T_1 images与3D DESS融合图在隐匿性膝关节软骨损伤中的应用

目的探讨T_2 star mapping、T_1 images与3D DESS融合伪彩图在关节软骨损伤中的诊断价值。方法对26例关节软骨损伤患者行T_2 star mapping、T_1 images和3D DESS扫描,并将T_1 images、T_2 star mapping与3D DESS图像融合,评价患者股骨、胫骨、髌骨关节软骨损伤程度并与关节镜结果对比,计算融合伪彩图诊断软骨损伤的特异性、敏感性及与关节镜诊断结果一致性。结果 T_1 images-3D DESS融合伪彩图诊断关节软骨损伤的敏感度、特异度及Kappa值分别为92.8%、93.0%、0.769,T_2 star mapping-3D DESS融合伪彩图诊断关节软骨损伤的敏感度、特异度及Kappa值分别为91.4%、94.2%、0.787。结论 T_2 star mapping、T_1 images与3D DESS融合伪彩图在关节软骨早期损伤评价上优于关节镜。

联合多时相GF-6 WFV和Sentinel-2的森林类型识别

【目的】我国南方地区多云雨,地型较破碎,森林类型精细识别较为困难,探讨联合多源、多时相的遥感数据对森林类型识别具有重要意义。【方法】以江西省信丰县为研究区,基于2019年森林资源二类调查数据,将森林划分为松林、杉木林、阔叶林、针叶混交林、针阔混交林、竹林、灌木林和其他林地等8种类型,利用随机森林算法比较GF-6 WFV和Sentinel-2最佳时相相同波段(紫/深蓝、蓝、绿、红、近红外、红边)和不同波段(黄边、短波红外)的森林类型识别能力,构建联合光谱特征集。联合多时相GF-6 WFV和Sentinel-2,构建多时相植被指数特征集,结合联合光谱特征集、纹理特征和地形特征,通过随机森林和递归消除法构建特征变量优选数据集进行森林类型识别,利用混淆矩阵和森林类型的实际分布对识别结果进行精度验证。【结果】(1)GF-6 WFV蓝、绿和红波段组合的总体精度为58.31%,分别加入紫、近红外、红边、黄边和Sentinel-2短波红外波段后,其总体精度分别提高1.99%、8.90%、10.71%、1.50%和14.10%;Sentinel-2蓝、绿和红波段组合的总体精度为54.68%,分别加入深蓝、近红外、红边、短波红外和GF-6 WFV黄边波段后,其总体精度分别提高3.30%、10.82%、12.92%、17.31%和3.97%。(2)特征变量优选数据集的总体精度和Kappa系数为80.80%和75.56%,贡献程度大小依次为GF-6 WFV多时相植被指数、Sentinel-2多时相植被指数、GF-6 WFV光谱特征、Sentinel-2光谱特征、地形特征和纹理特征,贡献率分别为40.44%、23.23%、18.12%、10.21%、4.61%和3.39%。(3)松林、杉木林、阔叶林、针叶混交林、针阔混交林、竹林、灌木林和其他林地的制图精度分别为86.97%、85.60%、88.61%、9.43%、19.01%、53.60%、86.90%和82.56%,用户精度分别为81.42%、79.79%、77.57%、71.43%、81.82%、67.00%、87.74%和82.88%,识别结果与研究区实际森林类型分布较吻合。【结论】联合多时相GF-6 WFV和Sentinel-2可以综合多时相、多源影像的优点,能够有效提高森林类型的识别精度。

基于Image 2和岭回归模型估测肉牛体尺、体重

为了建立快速、经济、准确、可行的预测肉牛体尺、体重性状指标的方法,试验利用Image 2图像识别技术和岭回归模型预测肉牛的体尺、体重,并与实测值进行比较,经过交叉验证该模型估测结果的准确性。结果表明:试验肉牛预测体尺与真实体尺的相关性>40%,其预测平均偏差小于0.04m。利用岭回归模型预测体重,通过3倍的交叉验证获得93%以上的准确率。说明可以利用图像识别技术与岭回归模型直接预测肉牛体尺、体重。

IMAGE 2型糖尿病预防指南要点与点评

新近,欧洲糖尿病预防指南与培训标准工作组(Development and Implementation of a European Guideline and Training Standards for Diabetes Prevention,IMAGE)颁布了2型糖尿病(T2DM)预防指南,其要点摘译如下:

基于GF-1卫星遥感反演排水河沟水体溶存N_(2)O浓度模型对比研究

[目的]探究利用GF-1卫星数据反演水体溶存氧化亚氮(N_(2)O)浓度的可行性,为实现低成本、高效率的水质实时监测提供有效途径。[方法]以宁夏青铜峡灌区第1和第5排水河沟为研究对象,选取与排水河沟水体溶存N_(2)O浓度相关性高的GF-1卫星影像波段反射率和水质参数作为自变量,通过最优子集筛选法确定最优自变量组合,分别构建了多元线性回归、BP神经网络和支持向量机模型,对水体溶存N_(2)O浓度进行预测对比。[结果]水温(T)、溶解性有机碳(DOC)等是影响水体溶存N_(2)O浓度的主要因素,同时近红外(NIR)等卫星波段与水体溶存N_(2)O浓度变化趋势显著相关。当自变量包括T,NIR等7个因素时,模型预测效果最佳。在3种模型中,BP神经网络模型验证结果R^(2)为0.64,具有最高预测精度。[结论]GF-1卫星数据以及水质参数与水体溶存N_(2)O浓度存在复杂的相关性关系,且BP神经网络能够实现利用GF-1卫星数据较高精度地反演水体溶存N_(2)O浓度。

级联语义分割和边缘检测的GF-2影像耕地提取

针对山地丘陵区的坡耕地和小面积耕地碎片边界模糊不清、分类提取困难的问题,以GF-2影像为数据源,提出了一种级联语义分割和边缘检测模型的遥感影像耕地信息提取方法。首先,针对不同类型耕地的特点选择级联方式;其次,将耕地边缘作为独立的特征类别,结合改进U-Net、DeeplabV3+和DexiNed模型,融合面特征和线特征,使得耕地边缘特征与语义特征能够进行互补,从而提高耕地提取的准确性,实现对复杂地形背景噪声的抑制和不同类型耕地的提取。实验结果表明,对比单一模型DeeplabV3+和U-Net,级联模型的耕地信息提取的总体精度、Kappa系数和F1值均有大幅度提升,针对不同类型耕地级联模型提取的耕地结果更接近真实耕地标注,漏提、误提区域远低于单一模型。

基于GF-2遥感影像和改进后PSPNet模型的丘陵地区耕地图斑提取方法

针对丘陵地区耕地地块具有边界模糊、覆盖物种类多样、大小和空间位置分布不规则等特点,传统分类方法难以快速准确提取耕地信息的问题,本文以四川省金堂县竹篙镇和高板镇为研究区域,利用高分二号卫星影像和改进后的PSPNet语义分割网络模型进行耕地图斑提取研究。在模型训练中,引入CBAM注意力模块以提高整个网络的特征提取和表达能力,采用余弦退火学习率以加快模型的收敛速度。结果表明,改进后的PSPNet模型在丘陵地区耕地提取精度方面取得了显著提高,耕地识别精度达到了95.69%,比标准PSPNet模型提高了1.07%,比Unet++,DeepLabv3+和支持向量机方法方法提高了1.32%,1.75%和6.33%。基于改进后的PSPNet模型具有更强的特征提取和表达能力,可以更准确地提取丘陵地区的耕地信息,为农业决策提供更准确的数据支持,促进农业智能化和精准化,提高农作物产量和质量,推动农业现代化进程。

基于GF-2和ASTER数据青海德龙地区构造蚀变信息提取及找矿预测

德龙地区位于青海东昆仑金及多金属成矿带东段,是该成矿带内极具勘查潜力区域之一,受限于偏远的地理位置和崎岖的地形条件,大比例尺地球化学勘查及常规地质调查工作难以直接开展。为此,基于ASTER和GF-2数据,通过分析不同空间分辨率遥感影像的色调、几何结构和纹理特征进行线性和环形构造识别;同时,根据主要蚀变矿物光谱特征分析,利用ASTER可见光—近红外波段和短波红外波段,采用“掩模+主成分分析”方法提取铁化、Al-OH和Mg-OH蚀变信息;在此基础上,结合多元地学信息及野外调查结果,综合分析遥感解译构造蚀变信息与研究区金矿化的内在联系,建立基于区内金矿床的遥感找矿预测模型,并以此为依据划分出找矿远景区3处。通过野外查证,在德龙找矿远景区新发现金矿体多条。研究结果表明,融合遥感数据和地理信息系统技术可有效识别地表热液蚀变和构造空间结构特征,能够为该地区进一步找矿预测提供参考和依据。

基于GF-2影像的武汉市九峰山国家森林公园地上碳储量估算

探究国产高分辨率数据在森林碳储量估算研究中的潜力,为构建森林碳储量估算模型提供新思路。选取武汉市九峰山国家森林公园为研究对象,以GF-2遥感影像为数据源,结合地面实测信息,对研究区森林地上碳储量进行估算,共提取6个植被指数、4个波段值、8种纹理特征,筛选出9个与实测碳储量相关的遥感变量,运用线性与非线性方程对单个高相关变量和多个相关变量进行建模,选出最优模型,为进一步提高预测精度,将模型代入4种纹理窗口(3×3、5×5、7×7、9×9)。结果表明:通过遥感图像提取的植被指数之间,具有较强的共线性,单变量建立的模型精度低于多变量模型;利用均方根误差RMSE与决定系数R^(2)对4个窗口下模型的预测精度进行评价,模型在5×5窗口下预测效果最好(R^(2)=0.73,RMSE=0.5),3×3窗口下预测效果最差(R^(2)=0.64,RMSE=0.8),将所有估测模型进行比较,在纹理窗口下模型精度提高了0.11。利用5×5窗口下构建的多变量模型对研究区碳储量进行估算,九峰山国家森林公园碳储总量为1.06×10^(4) t,总体平均碳密度为84.59 t/hm^(2),具有一定的固碳作用。选用国产高分辨率影像GF-2数据对武汉市九峰山森林公园进行反演研究,能很好地运用在森林植被碳储量定量与生长状况领域。研究结果对“双碳”目标下森林生态系统碳汇监测与管理具有重要科学意义。

T2-weighted imaging-based radiomic-clinical machine learning model for predicting the differentiation of colorectal adenocarcinoma

BACKGROUND The study on predicting the differentiation grade of colorectal cancer(CRC)based on magnetic resonance imaging(MRI)has not been reported yet.Developing a non-invasive model to predict the differentiation grade of CRC is of great value.AIM To develop and validate machine learning-based models for predicting the differ-entiation grade of CRC based on T2-weighted images(T2WI).METHODS We retrospectively collected the preoperative imaging and clinical data of 315 patients with CRC who underwent surgery from March 2018 to July 2023.Patients were randomly assigned to a training cohort(n=220)or a validation cohort(n=95)at a 7:3 ratio.Lesions were delineated layer by layer on high-resolution T2WI.Least absolute shrinkage and selection operator regression was applied to screen for radiomic features.Radiomics and clinical models were constructed using the multilayer perceptron(MLP)algorithm.These radiomic features and clinically relevant variables(selected based on a significance level of P<0.05 in the training set)were used to construct radiomics-clinical models.The performance of the three models(clinical,radiomic,and radiomic-clinical model)were evaluated using the area under the curve(AUC),calibration curve and decision curve analysis(DCA).RESULTS After feature selection,eight radiomic features were retained from the initial 1781 features to construct the radiomic model.Eight different classifiers,including logistic regression,support vector machine,k-nearest neighbours,random forest,extreme trees,extreme gradient boosting,light gradient boosting machine,and MLP,were used to construct the model,with MLP demonstrating the best diagnostic performance.The AUC of the radiomic-clinical model was 0.862(95%CI:0.796-0.927)in the training cohort and 0.761(95%CI:0.635-0.887)in the validation cohort.The AUC for the radiomic model was 0.796(95%CI:0.723-0.869)in the training cohort and 0.735(95%CI:0.604-0.866)in the validation cohort.The clinical model achieved an AUC of 0.751(95%CI:0.661-0.842)in the training cohort and 0.676(95%CI:0.525-0.827)in the validation cohort.All three models demonstrated good accuracy.In the training cohort,the AUC of the radiomic-clinical model was significantly greater than that of the clinical model(P=0.005)and the radiomic model(P=0.016).DCA confirmed the clinical practicality of incorporating radiomic features into the diagnostic process.CONCLUSION In this study,we successfully developed and validated a T2WI-based machine learning model as an auxiliary tool for the preoperative differentiation between well/moderately and poorly differentiated CRC.This novel approach may assist clinicians in personalizing treatment strategies for patients and improving treatment efficacy.

血清TGF-β1、Nesfatin-1、SFRP5与2型糖尿病胰岛素抵抗关系探究

目的:探讨血清转化生长因子-β1(TGF-β1)、Nesfatin-1、分泌型卷曲相关蛋白-5(SFRP5)与2型糖尿病胰岛素抵抗的关系。方法:选取我院2021年4月-2023年4月收治的148例2型糖尿病患者为研究对象,根据稳态模型胰岛素抵抗指数(HOMA-IR)检测结果将患者分为轻度胰岛素抵抗组(91例,HOMA-IR<2.5)和中重度胰岛素抵抗组(57例,HOMA-IR≥2.5)。比较两组入院时血清TGF-β1、Nesfatin-1、SFRP5水平,分析入院时血清TGF-β1、Nesfatin-1、SFRP5水平与HOMA-IR的相关性,分析发生中重度胰岛素抵抗的影响因素、联合检测的诊断价值及危险度。结果:入院时,中重度胰岛素抵抗组血清TGF-β1、Nesfatin-1水平高于轻度胰岛素抵抗组,SFRP5水平低于轻度胰岛素抵抗组(P<0.05);入院时,HOMA-IR与血清TGF-β1、Nesfatin-1水平呈正相关,与SFRP5水平呈负相关(P<0.05);入院时,血清TGF-β1、Nesfatin-1、SFRP5水平为2型糖尿病患者发生中重度胰岛素抵抗的影响因素(P<0.05);入院时血清TGF-β1、Nesfatin-1、SFRP5水平联合诊断中重度胰岛素抵抗的AUC为0.808,最佳诊断敏感度为91.23%,特异度为70.33%;入院时血清TGF-β1、Nesfatin-1、SFRP5高水平患者发生中重度胰岛素抵抗的危险度是低水平的3.560、2.134、0.341倍(P<0.05)。结论:血清TGF-β1、Nesfatin-1、SFRP5水平与2型糖尿病胰岛素抵抗密切相关,可作为临床诊断病情的参考依据。

Performances of conventional fusion methods evaluated for inland water body observation using GF-1 image

Satellite remote sensing of inland water body requires a high spatial resolution and a multiband narrow spectral resolution, which makes the fusion between panchromatic(PAN) and multi-spectral(MS) images particularly important. Taking the Daquekou section of the Qiantang River as an observation target, four conventional fusion methods widely accepted in satellite image processing, including pan sharpening(PS), principal component analysis(PCA), Gram-Schmidt(GS), and wavelet fusion(WF), are utilized to fuse MS and PAN images of GF-1.The results of subjective and objective evaluation methods application indicate that GS performs the best,followed by the PCA, the WF and the PS in the order of descending. The existence of a large area of the water body is a dominant factor impacting the fusion performance. Meanwhile, the ability of retaining spatial and spectral informations is an important factor affecting the fusion performance of different fusion methods. The fundamental difference of reflectivity information acquisition between water and land is the reason for the failure of conventional fusion methods for land observation such as the PS to be used in the presence of the large water body. It is suggested that the adoption of the conventional fusion methods in the observing water body as the main target should be taken with caution. The performances of the fusion methods need re-assessment when the large-scale water body is present in the remote sensing image or when the research aims for the water body observation.

Infrared image segmentation method based on 2D histogram shape modification and optimal objective function

In the methods of image thresholding segmentation, such methods based on two-dimensional （2D） histogram and optimal objective functions are important. However, when they are used for infrared image segmentation, they are weak in suppressing background noises and worse in segmenting targets with non-uniform gray level. The concept of 2D histogram shape modification is proposed, which is realized by target information prior restraint after enhancing target information using plateau histogram equalization. The formula of 2D minimum Renyi entropy is deduced for image segmentation, then the shape-modified 2D histogram is combined wfth four optimal objective functions （i.e., maximum between-class variance, maximum entropy, maximum correlation and minimum Renyi entropy） respectively for the appli- cation of infrared image segmentation. Simultaneously, F-measure is introduced to evaluate the segmentation effects objectively. The experimental results show that F-measure is an effective evaluation index for image segmentation since its value is fully consistent with the subjective evaluation, and after 2D histogram shape modification, the methods of optimal objective functions can overcome their original forms＇ deficiency and their segmentation effects are more or less improvements, where the best one is the maximum entropy method based on 2D histogram shape modification.

Defect detection method based on 2D entropy image segmentation

In order to improve the work efficiency of non-destructive testing(NDT)and the reliability of NDT results,an automatic method to detect defects in the ultrasonic image was researched.According to the characterization of ultrasonic D-scan image,clutter wave suppression and de-noising were presented firstly.Then,the image is processed by binaryzation using KSW 2 D entropy based on image segmentation method.The results showed that,the global threshold based segmentation method was somewhat ineffective for D-scan image because of under-segmentation.Especially,when the image is big in size,small targets which are composed by a small amount of pixels are often undetected.Whereas,local threshold based image segmentation method is effective in recognizing small defects because it takes local image character into account.

Sentiment Analysis Using E-Commerce Review Keyword-Generated Image with a Hybrid Machine Learning-Based Model

In the context of the accelerated pace of daily life and the development of e-commerce,online shopping is a mainstreamway for consumers to access products and services.To understand their emotional expressions in facing different shopping experience scenarios,this paper presents a sentiment analysis method that combines the ecommerce reviewkeyword-generated imagewith a hybrid machine learning-basedmodel,inwhich theWord2Vec-TextRank is used to extract keywords that act as the inputs for generating the related images by generative Artificial Intelligence(AI).Subsequently,a hybrid Convolutional Neural Network and Support Vector Machine(CNNSVM)model is applied for sentiment classification of those keyword-generated images.For method validation,the data randomly comprised of 5000 reviews from Amazon have been analyzed.With superior keyword extraction capability,the proposedmethod achieves impressive results on sentiment classification with a remarkable accuracy of up to 97.13%.Such performance demonstrates its advantages by using the text-to-image approach,providing a unique perspective for sentiment analysis in the e-commerce review data compared to the existing works.Thus,the proposed method enhances the reliability and insights of customer feedback surveys,which would also establish a novel direction in similar cases,such as social media monitoring and market trend research.

2-D mini mumfuzzy entropy method of image thresholding based on genetic algorithm

A new image thresholding method is introduced, which is based on 2-D histgram and minimizing the measures of fuzziness of an input image. A new definition of fuzzy membership function is proposed, it denotes the characteristic relationship between the gray level of each pixel and the average value of its neighborhood. When the threshold is not located at the obvious and deep valley of the histgram, genetic algorithm is devoted to the problem of selecting the appropriate threshold value. The experimental results indicate that the proposed method has good performance.

Mapping soil organic matter in cultivated land based on multi-year composite images on monthly time scales

Rapid and accurate acquisition of soil organic matter(SOM)information in cultivated land is important for sustainable agricultural development and carbon balance management.This study proposed a novel approach to predict SOM with high accuracy using multiyear synthetic remote sensing variables on a monthly scale.We obtained 12 monthly synthetic Sentinel-2 images covering the study area from 2016 to 2021 through the Google Earth Engine(GEE)platform,and reflectance bands and vegetation indices were extracted from these composite images.Then the random forest(RF),support vector machine(SVM)and gradient boosting regression tree(GBRT)models were tested to investigate the difference in SOM prediction accuracy under different combinations of monthly synthetic variables.Results showed that firstly,all monthly synthetic spectral bands of Sentinel-2 showed a significant correlation with SOM(P<0.05)for the months of January,March,April,October,and November.Secondly,in terms of single-monthly composite variables,the prediction accuracy was relatively poor,with the highest R^(2)value of 0.36 being observed in January.When monthly synthetic environmental variables were grouped in accordance with the four quarters of the year,the first quarter and the fourth quarter showed good performance,and any combination of three quarters was similar in estimation accuracy.The overall best performance was observed when all monthly synthetic variables were incorporated into the models.Thirdly,among the three models compared,the RF model was consistently more accurate than the SVM and GBRT models,achieving an R^(2)value of 0.56.Except for band 12 in December,the importance of the remaining bands did not exhibit significant differences.This research offers a new attempt to map SOM with high accuracy and fine spatial resolution based on monthly synthetic Sentinel-2 images.