Hand segmentation from a single depth image based on histogram threshold selection and shallow CNN

Real-time hand gesture recognition technology significantly improves the user's experience for virtual reality/augmented reality(VR/AR) applications, which relies on the identification of the orientation of the hand in captured images or videos. A new three-stage pipeline approach for fast and accurate hand segmentation for the hand from a single depth image is proposed. Firstly, a depth frame is segmented into several regions by histogrambased threshold selection algorithm and by tracing the exterior boundaries of objects after thresholding. Secondly, each segmentation proposal is evaluated by a three-layers shallow convolutional neural network(CNN) to determine whether or not the boundary is associated with the hand. Finally, all hand components are merged as the hand segmentation result. Compared with algorithms based on random decision forest(RDF), the experimental results demonstrate that the approach achieves better performance with high-accuracy(88.34% mean intersection over union, mIoU) and a shorter processing time(≤8 ms).

Threshold Selection Method Based on Reciprocal Gray Entropy and Artificial Bee Colony Optimization

Since the logarithmic form of Shannon entropy has the drawback of undefined value at zero points,and most existing threshold selection methods only depend on the probability information,ignoring the within-class uniformity of gray level,a method of reciprocal gray entropy threshold selection is proposed based on two-dimensional(2-D)histogram region oblique division and artificial bee colony(ABC)optimization.Firstly,the definition of reciprocal gray entropy is introduced.Then on the basis of one-dimensional(1-D)method,2-D threshold selection criterion function based on reciprocal gray entropy with histogram oblique division is derived.To accelerate the progress of searching the optimal threshold,the recently proposed ABC optimization algorithm is adopted.The proposed method not only avoids the undefined value points in Shannon entropy,but also achieves high accuracy and anti-noise performance due to reasonable 2-D histogram region division and the consideration of within-class uniformity of gray level.A large number of experimental results show that,compared with the maximum Shannon entropy method with 2-D histogram oblique division and the reciprocal entropy method with 2-D histogram oblique division based on niche chaotic mutation particle swarm optimization(NCPSO),the proposed method can achieve better segmentation results and can satisfy the requirement of real-time processing.

An Effective Method of Threshold Selection for Small Object Image

The image segmentation difficulties of small objects which are much smaller than their background often occur in target detection and recognition. The existing threshold segmentation methods almost fail under the circumstances. Thus, a threshold selection method is proposed on the basis of area difference between background and object and intra-class variance. The threshold selection formulae based on one-dimensional (1-D) histogram, two-dimensional (2-D) histogram vertical segmentation and 2-D histogram oblique segmentation are given. A fast recursive algorithm of threshold selection in 2-D histogram oblique segmentation is derived. The segmented images and processing time of the proposed method are given in experiments. It is compared with some fast algorithms, such as Otsu, maximum entropy and Fisher threshold selection methods. The experimental results show that the proposed method can effectively segment the small object images and has better anti-noise property.

Performance Analysis of a Threshold-Based Relay Selection Algorithm in Wireless Networks

Relay selection is an effective method to realize the cooperative diversity gain in wireless networks. In this paper, we study a threshold-based single relay selection algorithm. A reasonable threshold value is set at each relay node, and the first relay with the instantaneous channel gain larger than the threshold will be se-lected to cooperate with the source. The exact and closed form expression for its outage probability is de-rived over independent, non-identically distributed (i. n. i. d) Rayleigh channels. The complexity of the algo-rithm is also analyzed in detail. Simulation results are presented to verify our theoretical analysis.

Threshold Selection Study on Fisher Discriminant Analysis Used in Exon Prediction for Unbalanced Data Sets

In gene prediction, the Fisher discriminant analysis (FDA) is used to separate protein coding region (exon) from non-coding regions (intron). Usually, the positive data set and the negative data set are of the same size if the number of the data is big enough. But for some situations the data are not sufficient or not equal, the threshold used in FDA may have important influence on prediction results. This paper presents a study on the selection of the threshold. The eigen value of each exon/intron sequence is computed using the Z-curve method with 69 variables. The experiments results suggest that the size and the standard deviation of the data sets and the threshold are the three key elements to be taken into consideration to improve the prediction results.

Threshold Selection and Resource Allocation for Quantized Identification

This paper is concerned with the optimal threshold selection and resource allocation problems of quantized identification, whose aims are improving identification efficiency under limited resources. Firstly, the first-order asymptotically optimal quantized identification theory is extended to the weak persistent excitation condition. Secondly, the characteristics of time and space complexities are established based on the Cramér-Rao lower bound of quantized systems. On these basis, the optimal selection methods of fixed thresholds and adaptive thresholds are established under aperiodic signals, which answer how to achieve the best efficiency of quantized identification under the same time and space complexity. In addition, based on the principle of maximizing the identification efficiency under a given resource, the optimal resource allocation methods of quantized identification are given for the cases of fixed thresholds and adaptive thresholds, respectively, which show how to balance time and space complexity to realize the best identification efficiency of quantized identification.

基于多源约束自适应视觉SLAM关键帧选取研究

该文针对现有关键帧选择方法在复杂场景下的稳定性和适应性方面不足问题,提出一种多源约束的自适应视觉SLAM关键帧选取方法。该算法基于相机几何测量原理,设计自适应阈值进行关键帧选取策略;针对复杂环境下的剧烈运动情况,设计基于IMU的实时状态检测机制和熵函数约束标准,进一步提高关键帧选取的稳定性和适应性。在EuRoC数据集和TUM数据集上对该方法进行定性和定量评估。在单目惯性和立体惯性模式下,将估计轨迹与参考轨迹进行对比,以绝对轨迹误差(absolute trajectory error,ATE)、关键帧数量和运行时间作为评判指标,并与ORB-SLAM3方法进行比较。结果显示,提出的方法可显著提高视觉SLAM在复杂环境下的定位精度和稳定性。

基于大津阈值的量子图像分割方法

为解决量子计算机上图像分割的问题,研究一种基于大津阈值的量子图像分割方法。设计量子加法器、量子减法器、量子乘法器、量子除法器和直方图的基本模块,在此基础上,设计累积直方图、累积平均值、类间方差等子模块,通过对每个灰度级所对应的类间方差进行排序,选取所有类间方差的最大值作为大津阈值,对得到的大津阈值进行图像二值化量子线路设计。通过在经典计算机上的仿真验证了方法的执行效果,基于所用基本量子门数量,分析量子线路的复杂度,其结果表明,所提方法可以实现对经典方法的加速。

一种基于主动学习的开放集图像识别方法

开放集识别(Open Set Recognition,OSR)的主要目的是识别未标记数据中的新类样本,同时对已见类样本进行正确分类.现有的大多数识别方法对未标记数据的评估和伪标记信息的利用不足.本文提出一种基于主动学习的开放集图像识别方法(Open Set Image Recognition Method Based on Active Learning,AC-OSIR),充分利用未标记数据提升开放集识别性能.通过引入已见类别的语义知识,构建语义知识和图像特征的映射关系.对于未标记数据,利用阈值选择策略区分开放集样本和已见类样本,通过主动学习模型迭代地识别高置信度开放集样本和已见类样本,并将高置信度已见类样本添加到标记数据集中.本文在图像分类数据集CIFAR-10、TIN和LSUN,以及两个合成数据集的实验结果表明了基于主动学习的开放集图像识别方法的有效性.

一种多尺度图像融合的冷冻电镜颗粒挑选方法

当前主流的冷冻电镜颗粒挑选方法往往需要大量人工生成的训练集或者优质颗粒模板,或者颗粒挑选过程极为复杂。为了提高冷冻电镜颗粒挑选的效率,简化颗粒挑选流程,提出一种自动挑选颗粒方法,在图像预处理阶段使用基于Lanczos采样图像融合方法提高图像质量,随后使用基于最大类间方差的图像阈值分割方法分离颗粒与背景,实现颗粒挑选。在EMPAIR公共数据集的实验结果表明,该方法与其他方法相比,具有更高的召回率与精确率。

Spatial pattern recognition for near-surface high temperature increases in mountain areas using MODIS and SRTM DEM

Abrupt near-surface temperature changes in mountainous areas are a special component of the mountain climate system.Fast and accurate measurements of the locations,intensity,and width of the near-surface changes are necessary but highly difficult due to the complicated environmental conditions and instrumental issues.This paper develops a spatial pattern recognition method to measure the near-surface high temperature increase(NSHTI),one of the lesser-attended changes.First,raster window measurement was proposed to calculate the temperature lapse rate using MODIS land surface temperature and SRTM DEM data.It fully considers the terrain heights of two neighboring cells on opposite or adjacent slopes with a moving window of 3×3 cell size.Second,a threshold selection was performed to identify the NSHTI cells using a threshold of-0.65℃/100 m.Then,the NSHTI strips were parameterized through raster vectorization and spatial analysis.Taking Yunnan,a mountainous province in southwestern China,as the study area,the results indicate that the NSHTI cells concentrate in a strip-like pattern along the mountains and valleys,and the strips are almost parallel to the altitude contours with a slight northward uplift.Also,they are located mostly at a 3/5 height of high mountains or within 400 m from the valley floors,where the controlling topographic index is the altitude of the terrain trend surface but not the absolute elevation and the topographic uplift height and cutting depth.Additionally,the NSHTI intensity varies with the geographic locations and the proportions increase with an exponential trend,and the horizontal width has a mean of about 1000 m and a maximum of over 5000 m.The result demonstrates that the proposed method can effectively recognize NSHTI boundaries over mountains,providing support for the modeling of weather and climate systems and the development of mountain resources.

Improving Video Watermarking through Galois Field GF(2^(4)) Multiplication Tables with Diverse Irreducible Polynomials and Adaptive Techniques

Video watermarking plays a crucial role in protecting intellectual property rights and ensuring content authenticity.This study delves into the integration of Galois Field(GF)multiplication tables,especially GF(2^(4)),and their interaction with distinct irreducible polynomials.The primary aim is to enhance watermarking techniques for achieving imperceptibility,robustness,and efficient execution time.The research employs scene selection and adaptive thresholding techniques to streamline the watermarking process.Scene selection is used strategically to embed watermarks in the most vital frames of the video,while adaptive thresholding methods ensure that the watermarking process adheres to imperceptibility criteria,maintaining the video's visual quality.Concurrently,careful consideration is given to execution time,crucial in real-world scenarios,to balance efficiency and efficacy.The Peak Signal-to-Noise Ratio(PSNR)serves as a pivotal metric to gauge the watermark's imperceptibility and video quality.The study explores various irreducible polynomials,navigating the trade-offs between computational efficiency and watermark imperceptibility.In parallel,the study pays careful attention to the execution time,a paramount consideration in real-world scenarios,to strike a balance between efficiency and efficacy.This comprehensive analysis provides valuable insights into the interplay of GF multiplication tables,diverse irreducible polynomials,scene selection,adaptive thresholding,imperceptibility,and execution time.The evaluation of the proposed algorithm's robustness was conducted using PSNR and NC metrics,and it was subjected to assessment under the impact of five distinct attack scenarios.These findings contribute to the development of watermarking strategies that balance imperceptibility,robustness,and processing efficiency,enhancing the field's practicality and effectiveness.

基于多方嵌入的逐步实体对齐方法

大多数实体对齐方法对知识图谱信息的利用不够充分,没有考虑实体间的互相选择,忽略了现实生活中很多实体在对方知识图谱中不存在等价实体的事实。针对以上问题,提出一种基于多方嵌入的逐步实体对齐方法。该方法对三元组信息、邻域信息、实体名称的语义信息和字符串信息进行多方嵌入生成相似度矩阵,再通过所提出的逐步实体对齐算法将目前彼此最为相似且相似度大于最小相似度阈值的两个实体进行匹配,直到剩余所有实体的相似度都不大于最小相似度阈值时停止匹配,在确保等价实体准确匹配的前提下,减小不存在等价实体时发生错误匹配的概率。在DBP15K数据集上进行了三项实验,结果证明了该方法和逐步实体对齐算法的有效性,以及多方嵌入中每个模块的必要性。

基于最小二乘支持向量机的造纸工控网络高隐蔽性入侵检测

造纸工控网络的数据特征具有复杂性和多样性,对于高隐蔽性入侵行为,其特征可能被混杂在正常操作和噪声中,增加了检测的难度。为此,提出基于最小二乘支持向量机的造纸工控网络高隐蔽性入侵检测方法。首先,使用CEEMD算法对网络数据进行分解,得到一系列固有模态分量(IMF),利用排列熵对IMF分量进行分析,确定高噪声IMF分量;使用小波降噪对高噪声IMF分量展开抗干扰处理。然后,使用互信息特征选择方法对抗干扰处理后的入侵数据进行特征提取。最后,将提取到的入侵数据特征作为输入数据,通过最小二乘支持向量机(LS-SVM)建立一个判别函数,该函数根据输入数据的特征值进行分类,并判断网络中是否存在高隐蔽性入侵行为。实验结果表明,所提方法最高入侵检测准确率达到0.98,Kappa统计量最大为0.99,表明所提方法的数据处理效果好、网络入侵检测精度高。

基于旋转框表示的光学遥感图像目标检测

旋转目标检测是遥感图像智能解译的关键步骤,然而遥感图像背景复杂、目标方向具有任意性、尺度差异大,实现准确的旋转目标检测有一定的困难。提出的ERDet结合显示视觉中心,提取遥感图像的全局信息与局部信息,结合自适应阈值样本选择的水平目标检测算法和长边定义法,预测遥感图像目标的类别、位置和旋转角度。在DOTA-v1.0数据集上的实验表明,该方法能够对不同尺度和方向的目标进行准确提取,实现了对遥感目标的精准检测。

基于最优阈值选择的L波段测风雷达杂波抑制方法

针对L波段测风雷达杂波尖峰干扰,使得测风雷达信号表现出奇异特性的问题,提出了基于最优阈值选择的L波段测风雷达杂波抑制方法。设计浮动阈值子波杂波处理器,采用最佳阈值控制虚警概率参量。根据测风雷达杂波测量原理,计算杂波的测风雷达接收功率。跟踪测风雷达在分辨单元天线波束中的主轴俯仰角,消除测风雷达定位误差引起的噪声脉动,实现雷达杂波的精准秒级跟踪。建立二维杂波网格单元,结合最优阈值选择方法,提取L波段测风信息滤波后的低频系数,获取雷达测风信道的频率域,达到杂波抑制的目的。由外场试验结果可知,所研究方法能够有效抑制杂波,I、Q路信号波动幅度范围分别为0.5~1.5 dbm、-1.5~-0.5 dbm,与原始波形一致。

基于FY-33D/MERSI-Ⅱ和Terra/MODIS数据的森林火点提取算法——以我国西南地区为例

中国西南地区森林覆盖率高,土地利用类型多样,人类活动频繁,森林火灾风险较高,防火任务十分艰巨。遥感技术具有覆盖范围广和周期性监测等优势,被广泛应用于森林火灾监测。气象卫星搭载了对火灾敏感的红外谱段传感器且时间分辨率较高,已成为森林火灾监测的重要遥感数据源。物理阈值法是遥感数据火点提取中比较成熟的算法,但其精度受所选阈值的影响较大,导致不同区域的火点提取精度存在差异。以中国西南地区的重庆市为例,提出了一种采用FY-3D/MERSI-Ⅱ和Terra/MODIS极轨卫星数据作为数据源进行森林火点监测的算法。首先,通过分析前一年MOD14监测产品的历史数据,确定该地区的初定火点阈值;随后,消除云层、水体和太阳耀斑的干扰;最后,采用上下文背景阈值法进行火点提取。对2022年8月重庆市山火的分析表明,该算法的火点提取平均准确率超过了90%。研究结果表明,根据区域特点调整阈值可以提高该地区火点监测的准确率。

基于边缘与区域结合的光学遥感图像舰船分割方法

针对光学遥感图像中舰船目标分割精度低等问题,提出了一种基于边缘与区域结合的图像中舰船目标分割方法。首先,利用导向滤波改进Canny算法对待分割舰船遥感图像进行预处理,以保持边缘完整结构,得到舰船目标粗分割结果;然后,提出基于布谷鸟搜索优化的倒数灰度熵阈值选取方法进行舰船目标细分割;最后,引入形态学方法对分割结果进行修正,确保舰船目标的完整性。大量试验结果表明,该方法表现出较优的分割性能,为舰船检测识别领域提供了更好的分割方案。

一种适用于高维回归的多阶段稀疏变量选择算法

针对高维线性模型中微阵列数据的变量选择问题,尤其自变量数量远远大于样本数量时,提出一种多阶段变量选择算法。算法基于阈值化弹性网正则化方法,结合逐级多重假设检验,能在多阶段实现变量选择,并保证模型稀疏性和预测精度。模拟数据和实证研究结果表明,算法具有良好的有限样本性能,能够在保持预测精度的同时恢复真实模型,显著减少假阳性变量数量。

An Improved Double-Threshold Method Based on Gradient Histogram

This paper analyzes the characteristics of the output gradient histogram and shortages of several traditional automatic threshold methods in order to segment the gradient image better. Then an improved double-threshold method is proposed, which is combined with the method of maximum classes variance, estimating-area method and double-threshold method. This method can automatically select two different thresholds to segment gradient images. The computer simulation is performed on the traditional methods and this algorithm and proves that this method can get satisfying result. Key words gradient histogram image - threshold selection - double-threshold method - maximum classes variance method CLC number TP 391. 41 Foundation item: Supported by the National Nature Science Foundation of China (50099620) and the Project of Chenguang Plan in Wuhan (985003062)Biography: YANG Shen (1977-), female, Ph. D. candidate, research direction: multimedia information processing and network technology.