An Improved Soft Subspace Clustering Algorithm for Brain MR Image Segmentation

In recent years,the soft subspace clustering algorithm has shown good results for high-dimensional data,which can assign different weights to each cluster class and use weights to measure the contribution of each dimension in various features.The enhanced soft subspace clustering algorithm combines interclass separation and intraclass tightness information,which has strong results for image segmentation,but the clustering algorithm is vulnerable to noisy data and dependence on the initialized clustering center.However,the clustering algorithmis susceptible to the influence of noisydata and reliance on initializedclustering centers andfalls into a local optimum;the clustering effect is poor for brain MR images with unclear boundaries and noise effects.To address these problems,a soft subspace clustering algorithm for brain MR images based on genetic algorithm optimization is proposed,which combines the generalized noise technique,relaxes the equational weight constraint in the objective function as the boundary constraint,and uses a genetic algorithm as a method to optimize the initialized clustering center.The genetic algorithm finds the best clustering center and reduces the algorithm’s dependence on the initial clustering center.The experiment verifies the robustness of the algorithm,as well as the noise immunity in various ways and shows good results on the common dataset and the brain MR images provided by the Changshu First People’s Hospital with specific high accuracy for clinical medicine.

Application of U-Net and Optimized Clustering in Medical Image Segmentation:A Review

As a mainstream research direction in the field of image segmentation,medical image segmentation plays a key role in the quantification of lesions,three-dimensional reconstruction,region of interest extraction and so on.Compared with natural images,medical images have a variety of modes.Besides,the emphasis of information which is conveyed by images of different modes is quite different.Because it is time-consuming and inefficient to manually segment medical images only by professional and experienced doctors.Therefore,large quantities of automated medical image segmentation methods have been developed.However,until now,researchers have not developed a universal method for all types of medical image segmentation.This paper reviews the literature on segmentation techniques that have produced major breakthroughs in recent years.Among the large quantities of medical image segmentation methods,this paper mainly discusses two categories of medical image segmentation methods.One is the improved strategies based on traditional clustering method.The other is the research progress of the improved image segmentation network structure model based on U-Net.The power of technology proves that the performance of the deep learning-based method is significantly better than that of the traditional method.This paper discussed both advantages and disadvantages of different algorithms and detailed how these methods can be used for the segmentation of lesions or other organs and tissues,as well as possible technical trends for future work.

Improved Spectral Clustering Clothing Image Segmentation Algorithm Based on Sparrow Search Algorithm

In the process of clothing image researching,how to segment the clothing quickly and accurately and retain the clothing style details as much as possible is the basis of subsequent image analysis.Spectral clustering clothing image segmentation algorithm is a common method in the process of clothing image extraction.However,the traditional model requires high computing power and is easily affected by the initial center of clustering.It often falls into local optimization.Aiming at the above two points,an improved spectral clustering clothing image segmentation algorithm is proposed in this paper.The Nystrom approximation strategy is introduced into the spectral mapping process to reduce the computational complexity.In the clustering stage,this algorithm uses the global optimization advantage of the particle swarm optimization algorithm and selects the sparrow search algorithm to search the optimal initial clustering point,to effectively avoid the occurrence of local optimization.In the end,the effectiveness of this algorithm is verified on clothing images in each environment.

An Improved Lung Cancer Segmentation Based on Nature-Inspired Optimization Approaches

The distinction and precise identification of tumor nodules are crucial for timely lung cancer diagnosis andplanning intervention. This research work addresses the major issues pertaining to the field of medical imageprocessing while focusing on lung cancer Computed Tomography (CT) images. In this context, the paper proposesan improved lung cancer segmentation technique based on the strengths of nature-inspired approaches. Thebetter resolution of CT is exploited to distinguish healthy subjects from those who have lung cancer. In thisprocess, the visual challenges of the K-means are addressed with the integration of four nature-inspired swarmintelligent techniques. The techniques experimented in this paper are K-means with Artificial Bee Colony (ABC),K-means with Cuckoo Search Algorithm (CSA), K-means with Particle Swarm Optimization (PSO), and Kmeanswith Firefly Algorithm (FFA). The testing and evaluation are performed on Early Lung Cancer ActionProgram (ELCAP) database. The simulation analysis is performed using lung cancer images set against metrics:precision, sensitivity, specificity, f-measure, accuracy,Matthews Correlation Coefficient (MCC), Jaccard, and Dice.The detailed evaluation shows that the K-means with Cuckoo Search Algorithm (CSA) significantly improved thequality of lung cancer segmentation in comparison to the other optimization approaches utilized for lung cancerimages. The results exhibit that the proposed approach (K-means with CSA) achieves precision, sensitivity, and Fmeasureof 0.942, 0.964, and 0.953, respectively, and an average accuracy of 93%. The experimental results prove thatK-meanswithABC,K-meanswith PSO,K-meanswith FFA, andK-meanswithCSAhave achieved an improvementof 10.8%, 13.38%, 13.93%, and 15.7%, respectively, for accuracy measure in comparison to K-means segmentationfor lung cancer images. Further, it is highlighted that the proposed K-means with CSA have achieved a significantimprovement in accuracy, hence can be utilized by researchers for improved segmentation processes of medicalimage datasets for identifying the targeted region of interest.

PHISHING WEB IMAGE SEGMENTATION BASED ON IMPROVING SPECTRAL CLUSTERING

This paper proposes a novel phishing web image segmentation algorithm which based on improving spectral clustering.Firstly,we construct a set of points which are composed of spatial location pixels and gray levels from a given image.Secondly,the data is clustered in spectral space of the similar matrix of the set points,in order to avoid the drawbacks of K-means algorithm in the conventional spectral clustering method that is sensitive to initial clustering centroids and convergence to local optimal solution,we introduce the clone operator,Cauthy mutation to enlarge the scale of clustering centers,quantum-inspired evolutionary algorithm to find the global optimal clustering centroids.Compared with phishing web image segmentation based on K-means,experimental results show that the segmentation performance of our method gains much improvement.Moreover,our method can convergence to global optimal solution and is better in accuracy of phishing web segmentation.

Clustering-Inverse: A Generalized Model for Pattern-Based Time Series Segmentation

Patterned-based time series segmentation (PTSS) is an important task for many time series data mining applications. In this paper, according to the characteristics of PTSS, a generalized model is proposed for PTSS. First, a new inter-pretation for PTSS is given by comparing this problem with the prototype-based clustering (PC). Then, a novel model, called clustering-inverse model (CI-model), is presented. Finally, two algorithms are presented to implement this model. Our experimental results on artificial and real-world time series demonstrate that the proposed algorithms are quite effective.

Agent Based Segmentation of the MRI Brain Using a Robust C-Means Algorithm

In the last decade, the MRI (Magnetic Resonance Imaging) image segmentation has become one of the most active research fields in the medical imaging domain. Because of the fuzzy nature of the MRI images, many researchers have adopted the fuzzy clustering approach to segment them. In this work, a fast and robust multi-agent system (MAS) for MRI segmentation of the brain is proposed. This system gets its robustness from a robust c-means algorithm (RFCM) and obtains its fastness from the beneficial properties of agents, such as autonomy, social ability and reactivity. To show the efficiency of the proposed method, we test it on a normal brain brought from the BrainWeb Simulated Brain Database. The experimental results are valuable in both robustness to noise and running times standpoints.

A method of automatic image segmentation for gas bubbles in water

This paper presents an algorithm of automatic bubble image segmentation using the improved ant colony optimization methodology. The ant colony optimization method is a metaheuristic algorithm, and has been applied in many fields. To reveal the versatility and appropriateness of automatic bubble image segmentation, the fuzzy clustering analysis method is employed in ant colony optimization algorithm. Compared with the well-known image feature extraction operators such as SUSAN and Canny, the proposed method can comparatively suitable to extract the gas bubbles image edge features. The experimental results show that the proposed method is effective and reliable, and can achieve satisfactory image edge extraction effect.

基于累积和事件段识别与改进谱聚类的锂离子电池储能系统内短路故障检测方法

锂离子电池系统的内短路故障可能导致严重安全事故,其检测受到在线检测实时性以及故障特征获得性制约,是当下锂离子电池储能系统安全运行亟待解决的问题。该文提出一种基于累积和(cumulative sum,CUSUM)事件段检测与改进谱聚类的锂离子电池储能系统内短路故障检测方法。首先,考虑内短路故障时的电压/温度变化特性,基于累积和事件突变点识别方法,识别疑似内短路故障事件段。其次,构建三维故障特征,刻画检测对象内短路故障特征属性。然后,构建基于Wasserstein测度的内短路故障特征距离矩阵,检测三维空间各点稀疏特性,客观划定故障聚类,实现内短路故障检测。搭建锂离子电池内短路实验平台、建立锂离子电池电–热耦合仿真模型,算例结果表明该文方法能够准确识别疑似内短路故障事件段,在不同串并联形式及故障类型下实现故障检测,证明了该文方法的正确性与可行性。

高精度视觉感应技术在水肥一体机中的应用

以农田水肥智能化管理为研究对象,构建了一种高精度视觉感应式水肥一体机。采用高精度视觉感应技术获取作物生长状态参数图像,基于超像素图像分割技术,对复杂的作物生长状态图像特征向量进行提取,采用模糊聚类算法对图像进行分割处理,根据目标图像的像素值统计结果进行生长状态预测,并结合环境参数信息,构建灌溉过程土壤电导率EC和pH预测模型。测试结果表明:水肥一体机控制系统能够有效预测作物对水肥需求,提高了灌溉过程混肥精度,可节约灌溉用水量、提升生产效率、降低人工成本。

基于激光点云的橡胶树参数反演与数字孪生构建

基于激光雷达(Light Detection And Ranging,LiDAR)数据重建树体三维模型并精准获取林木空间枝干结构参数对林木性状评价、森林动态经营管理与可视化研究具有重要意义。为此提出一种基于骨架细化提取的树木模型重建方法。首先,采用FocusS350/350 PLUS三维激光扫描仪获取3块不同树龄橡胶树的样地数据。然后,作为细化建模的重点,将枝干点云从原始树点中分离出来,再将其过度分割为若干点云簇,通过相邻点云簇判断是否有分枝以及动态确定骨架点间距,并将其运用在空间殖民算法以此来生成树的三维骨架点和骨架点连通性链表,根据连通链表结构自动识别树木中的主枝干和各个一级分枝,再通过广义圆柱体生成树干完成树木三维重建。最后,利用数字孪生技术对这3块不同树龄样地树木进行三维实景建模,使其穿越时空在同一空间中重现,以便更为直观地观察树木在生长过程中的形态变化。该算法得到的橡胶树胸径与实测值比对为,决定系数(R^(2))>0.91,均方根误差(root mean square Error,RMSE)<1.00 cm;主枝干与一级枝干的分枝角为,R^(2)>0.91,RMSE<2.93;一级枝干直径为,R^(2)>0.90,RMSE<1.41 cm;将3个树龄放在一起计算其生长参数,并与实测值进行对比,发现该算法同样适用于异龄林样地的各个生长参数计算。同时发现橡胶树的一级枝条的直径越大,其相对应的叶团簇体积就越大。运用人工智能的理论模型来处理林木的激光点云数据,旨在为森林的可视化以及树木骨架结构的智能化分析与处理等研究领域提供有价值的参考。

代理多目标粒子群驱动的粗糙聚类图像分割算法

为了提高粗糙聚类算法应用于图像分割时的分割效果,提出一种代理多目标粒子群驱动的粗糙聚类图像分割算法。首先,通过自适应确定粗糙聚类上、下近似的阈值,减少人为干预;其次,利用粗糙聚类中边界样本占比构建动态惩罚因子,进而结合聚类的紧致性和可分性度量构造粗糙聚类目标函数,并联合聚类的连通性函数从不同角度共同评价聚类质量;最后,设计代理辅助的精英多目标粒子群优化策略,筛选精英粒子更新种群,得到最终的聚类中心,从而避免粗糙聚类算法对初始中心敏感和易陷入局部最优的问题并提升优化效率。实验结果表明:所设计的优化策略在标准测试问题上能够得到更好的优化结果;对比其他图像分割算法,该算法分割效果最佳。

基于预训练模型的单帧航拍图像无监督语义分割

针对航拍图像语义分割成本高、通用性差和精度低等问题,提出了一种两阶段无监督语义分割网络(two-stage unsupervised semantic segmentation net, TUSSNet),针对单帧航拍图像训练进而生成最终的语义分割结果。算法分为2个阶段。首先,使用对比语言-图像预训练(contrastive language-image pretraining, CLIP)模型生成航拍图像的粗粒度语义标签,然后进行网络的预热训练。其次,在第一阶段的基础上,采用分割一切模型(segment anything model, SAM)对航拍图像进行细粒度类别预测,生成精细化类别掩码伪标签;然后迭代优化网络,得到最终语义分割结果。实验结果显示,相较于现有无监督语义分割方法,算法显著提高了航拍图像的分割精度,同时提供了准确的语义信息。

基于K-means算法的跨国零售商客户细分研究

随着经济全球化及大数据技术的蓬勃发展,跨国零售商之间的竞争日益激烈,根据客户特征进行客户细分,协助客户进行个性化的服务体验,有利于跨国零售商实现精准营销和高效的客户关系管理。为了提高客户细分的精度,本文提出一种基于RFM模型的K-means聚类算法,使用簇内误方差(SSE)和轮廓系数(Silhouette Coefficient)计算聚类个数,优化K值选取。本文选取一家跨国零售商的数据进行实证检验,对细分后的结果进行特征分析,将客户划分为核心型客户、维护型客户和风险型客户三种类别,并为不同客户群体提供差异化营销策略,仅供参考。

一种皮肤镜图像分割方法

随着皮肤镜技术的发展,获取的高分辨率皮肤镜图像较大,采用SLIC算法构建超像素可在后续采用Ncut聚类融合超像素时降低内存要求并减少运算时间。因此提出一种基于改进SLIC与Ncut聚类算法相结合的皮肤镜图像分割算法。实验结果表明,与对比算法相比,提出的改进算法在ISIC2018数据集中低对比度、亮度不均匀皮肤镜图像上的分割有更好的表现。

一种新的高光谱遥感图像超像素分割方法

为了解决简单线性迭代聚类算法在高光谱遥感图像超像素分割任务中分割精度较低的问题,提出一种基于多级线性迭代聚类结合改进标签传播算法(LPA)的新的无监督高光谱遥感图像超像素分割方法。首先,扩充简单线性迭代聚类(SLIC)的适用范围至多通道对高光谱图像进行超像素初分割;然后,对色彩标准差较大的超像素进行多级迭代细致分割,引入基于局部二进制模式的高光谱遥感图像纹理特征提取方法计算高光谱图像纹理特征并融合多段光谱特征计算超像素间相似度以构建带权图网络;最后,改进LPA社区发现方法进行超像素合并,将改进的标签传播算法运用于超像素合并可以得到更加稳定准确的超像素合并效果,提高超像素分割精度。将该方法与多种方法进行比较,结果表明,该方法对高光谱遥感图像的超像素分割结果更准确,超像素边缘更贴合真实地物边界,能有效改善高光谱遥感图像超像素分割中精度较低的问题。

用户特征聚类和ICSA-SVR台区负荷预测

为提高配电网负荷预测精度,提出一种将模糊C均值(FCM)聚类与改进乌鸦搜索算法(ICSA)优化支持向量回归机(SVR)相结合的低压台区负荷预测模型。利用FCM算法对台区用户用电特征进行提取和聚类,消除用电行为特性差异对预测精度的影响,并构建ICSA-SVR模型,对各类用户的用电负荷进行回归预测,进而叠加得到台区负荷预测结果。结果显示,台区内不同类型用户之间的用电特性差异较大,可分冬季单峰型、夏季单峰型和冬夏双峰型三类,各台区负荷呈现不同的季节性波动;该方法能够明显提升台区负荷预测精度,预测结果可对电力生产运营提供指导。

Cluster parallel rendering based on encoded mesh

Use of compressed mesh in parallel rendering architecture is still an unexplored area, the main challenge of which is to partition and sort the encoded mesh in compression-domain. This paper presents a mesh compression scheme PRMC (Parallel Rendering based Mesh Compression) supplying encoded meshes that can be partitioned and sorted in parallel rendering system even in encoded-domain. First, we segment the mesh into submeshes and clip the submeshes’ boundary into Runs, and then piecewise compress the submeshes and Runs respectively. With the help of several auxiliary index tables, compressed submeshes and Runs can serve as rendering primitives in parallel rendering system. Based on PRMC, we design and implement a parallel rendering architecture. Compared with uncompressed representation, experimental results showed that PRMC meshes applied in cluster parallel rendering system can dramatically reduce the communication requirement.

Entropy-like distance driven fuzzy clustering with local information constraints for image segmentation

To improve the anti-noise ability of fuzzy local information C-means clustering, a robust entropy-like distance driven fuzzy clustering with local information is proposed. This paper firstly uses Jensen-Shannon divergence to induce a symmetric entropy-like divergence. Then the root of entropy-like divergence is proved to be a distance measure, and it is applied to existing fuzzy C-means(FCM) clustering to obtain a new entropy-like divergence driven fuzzy clustering, meanwhile its convergence is strictly proved by Zangwill theorem. In the end, a robust fuzzy clustering by combing local information with entropy-like distance is constructed to segment image with noise. Experimental results show that the proposed algorithm has better segmentation accuracy and robustness against noise than existing state-of-the-art fuzzy clustering-related segmentation algorithm in the presence of noise.