Single-Channel Speech Enhancement Using Critical-Band Rate Scale Based Improved Multi-Band Spectral Subtraction

This paper addresses the problem of single-channel speech enhancement in the adverse environment. The critical-band rate scale based on improved multi-band spectral subtraction is investigated in this study for enhancement of single-channel speech. In this work, the whole speech spectrum is divided into different non-uniformly spaced frequency bands in accordance with the critical-band rate scale of the psycho-acoustic model and the spectral over-subtraction is carried-out separately in each band. In addition, for the estimation of the noise from each band, the adaptive noise estimation approach is used and does not require explicit speech silence detection. The noise is estimated and updated by adaptively smoothing the noisy signal power in each band. The smoothing parameter is controlled by a-posteriori signal-to-noise ratio (SNR). For the performance analysis of the proposed algorithm, the objective measures, such as, SNR, segmental SNR, and perceptual evaluations of the speech quality are conducted for the variety of noises at different levels of SNRs. The speech spectrogram and objective evaluations of the proposed algorithm are compared with other standard speech enhancement algorithms and proved that the musical structure of the remnant noise and background noise is better suppressed by the proposed algorithm.

（Dis）Economies of Scale in Business Software Systems Development and Enhancement Projects

In the software engineering literature, it is commonly believed that economies of scale do not occur in case of software Development and Enhancement Projects （D＆EP）. Their per-unit cost does not decrease but increase with the growth of such projects product size. Thus this is diseconomies of scale that occur in them. The significance of this phenomenon results from the fact that it is commonly considered to be one of the fundamental objective causes of their low effectiveness. This is of particular significance with regard to Business Software Systems （BSS） D＆EP characterized by exceptionally low effectiveness comparing to other software D＆EP. Thus the paper aims at answering the following two questions： （1） Do economies of scale really not occur in BSS D＆EP？ （2） If economies of scale may occur in BSS D＆EP, what factors are then promoting them？ These issues classify into economics problems of software engineering research and practice.

3D pore-scale modeling of nanofluids-enhanced oil recovery

The numerical modeling of oil displacement by nanofluid based on three-dimensional micromodel of cores with different permeability was carried out by the volume of fluid(VOF)method with experimentally measured values of interfacial tension,contact angle and viscosity.Water-based suspensions of SiO_(2) nanoparticles with a concentration of 0–1%and different particle sizes were considered to study the effect of concentration and size of nanoparticles,displacement fluid flow rate,oil viscosity and core permeability on the efficiency of oil displacement by nanofluid.The oil recovery factor(ORF)increases with the increase of mass fraction of nanoparticles.An increase in nanoparticles’concentration to 0.5% allows an increase in ORF by about 19% compared to water flooding.The ORF increases with the decrease of nanoparticle size,and declines with the increase of displacing rate.It has been shown that the use of nanosuspensions for enhanced oil recovery is most effective for low-permeable reservoirs with highly viscous oil in injection modes with capillary number close to the immobilization threshold,and the magnitude of oil recovery enhancement decreases with the increase of displacement speed.The higher the oil viscosity,the lower the reservoir rock permeability,the higher the ORF improved by nanofluids will be.

A comprehensive review of medical image enhancement technologies

Image enhancement plays an important role in many applications of medical imaging. Image enhancement technologies can improve the qualities of medical images with low contrast and high level noise by stretching contrast, suppressing noise and so on. Such images processed by image enhancement technologies are helpful to doctors in analyses and diagnoses. In this paper, we present a technical review of various existing image enhancement methodologies which are often emoloved.

基于特征增强的双重注意力去雾网络

针对现有去雾方法处理的图像细节模糊和色彩偏差等问题,提出了一种基于特征增强的双重注意力去雾网络。该网络采用编码器-解码器结构,设计了一个双重注意力特征增强模块,其中,利用Ghost模块替代非线性卷积,实现模型轻量化处理,通过RFB充分融合不同尺度的特征,实现均匀去雾,引入双重注意力实现信息跨通道与空间交互,保证模型性能和抑制噪声特征。使用RESIDE数据集对网络进行训练和测试。实验结果表明,所提算法在主观视觉和客观评价指标上均有优异表现,能有效地提升网络的特征提取能力,实现对不同场景雾图的色彩恢复,增强图像的对比度和清晰度。

多尺度特征增强的街景绿色景观分割方法

针对街景图像中景观复杂多样且多种景观相互遮挡,绿色景观分割效果存在相似景观错分、边界分割模糊、细节丢失等问题,提出一种多尺度特征增强的城市绿色景观分割网络。在编码部分改进多尺度残差网络提取上下文信息以区分相似景观,同时构建多级特征聚合增强模块增强目标特征的边缘细节信息。增加双注意力机制,在局部特征上建模丰富的上下文联系。最后,将多级特征聚合增强模块同样引入解码器,并融合多层级特征来提高目标信息的恢复能力完善边缘信息。在公共街景数据集Cityscapes与自制数据集StreetData的消融实验表明,该网络与基础网络相比,平均交并比分别提高2.96%和5.57%。此外,在两个数据集上进行对比实验,该网络较对比模型平均交并比分别高1.25%~5.29%和1.52%~6.95%。定量分析与实验结果表明,该方法能够有效识别街景的绿色景观,实现高精度的城市绿色景观数据提取。

融合注意力和上下文信息的遥感图像小目标检测算法

对多尺度的遥感图像进行小目标检测时,基于深度学习的目标检测算法容易出现误检和漏检的情况。这是因为此类算法的特征提取模块进行了多次的下采样操作;而且未能根据不同类别、不同尺度的目标关注所需的上下文信息。为了解决该问题,提出一种融合注意力和上下文信息的遥感图像小目标检测算法ACM-YOLO(Attention-Context-Multiscale YOLO)。首先,应用细粒度的查询感知稀疏注意力以减少小目标特征信息的丢失,从而避免漏检;其次,设计局部上下文增强(LCE)函数以更好地关注不同类别的遥感目标所需的上下文信息,从而避免误检;最后,使用加权双向特征金字塔网络(BiFPN)强化特征融合模块对遥感图像小目标的多尺度特征融合能力,从而改善算法检测效果。在DOTA数据集和NWPU VHR-10数据集上进行对比实验和消融实验,以验证所提算法的有效性和泛化性。实验结果表明,在2个数据集上所提算法的平均精确率均值(mAP)分别达到了77.33%和96.12%,而相较于YOLOv5算法,召回率分别提升了10.00和7.50个百分点。可见,所提算法能有效提升mAP和召回率,减少误检和漏检。

微生物“清-防-驱”增效解堵体系的性能评价与现场试验

随着油田开发的不断深入,油井堵塞等储层伤害问题愈发严重,导致油井产量下降甚至停产。为有效提高石油开采效率,基于微生物菌体及其代谢产物的协同作用,选用具有解堵和驱油能力的菌株KC(Bacillus subtilis)、DY(Bacillus licheniformis)、DN1(Pseudomonas aeruginosa)及其代谢产物等进行复配而形成具有解堵、阻垢、缓蚀和驱油性能的3种增效解堵体系A、B和C。采用油水界面张力确定菌剂配比,通过垢物溶蚀、CaCO_(3)沉积、静态腐蚀、乳化指数测定和岩心驱替等室内实验进行体系性能评价,并将其用于现场试验。结果表明:增效解堵体系C对有机垢、无机垢和复合垢的溶蚀效率均在80%以上,阻垢率为95.50%,缓蚀率达61.44%,提高驱油效率14.58%。现场试验也证实体系C具有显著的增油效果,2口措施井平均单井日增油0.57 t。

Impacts of Multi-Scale Solar Activity on Climate.Part Ⅱ:Dominant Timescales in Decadal-Centennial Climate Variability

Part II of this study detects the dominant decadal-centennial timescales in four SST indices up to the 2010/2011 winter and tries to relate them to the observed 11-yr and 88-yr solar activity with the sunspot number up to Solar Cycle 24. To explore plausible solar origins of the observed decadal-centennial timescales in the SSTs and climate variability in general, we design a simple one-dimensional dynamical system forced by an annual cycle modulated by a small-amplitude single- or multi-scale ＂solar activity.＂ Results suggest that nonlinear harmonic and subharmonic resonance of the system to the forcing and period-doubling bifurcations are responsible for the dominant timescales in the system, including the 60-yr timescale that dominates the Atlantic Multidecadal Oscillation. The dominant timescales in the forced system depend on the system＇s parameter setting. Scale enhancement among the dominant response timescales may result in dramatic amplifications over a few decades and extreme values of the time series on various timescales. Three possible energy sources for such amplifications and extremes are proposed. Dynamical model results suggest that solar activity may play an important yet not well recognized role in the observed decadal-centennial climate variability. The atmospheric dynamical amplifying mechanism shown in Part I and the nonlinear resonant and bifurcation mechanisms shown in Part II help us to understand the solar source of the multi-scale climate change in the 20th century and the fact that different solar influenced dominant timescales for recurrent climate extremes for a given region or a parameter setting. Part II also indicates that solar influences on climate cannot be linearly compared with non-cyclic or sporadic thermal forcings because they cannot exert their influences on climate in the same way as the sun does.

Controlling factors of remaining oil distribution after water flooding and enhanced oil recovery methods for fracturecavity carbonate reservoirs in Tahe Oilfield

Based on comprehensive analysis of core, well logging, seismic and production data, the multi-scale reservoir space, reservoir types, spatial shape and distribution of fractures and caves, and the configuration relationship with production wells in fracture-cavity carbonate reservoirs were studied systematically, the influence of them on the distribution of residual oil was analyzed, and the main controlling factors mode of residual oil distribution after water flooding was established. Enhanced oil recovery methods were studied considering the development practice of Tahe oilfield. Research shows that the main controlling factors of residual oil distribution after water flooding in fracture-cavity carbonate reservoirs can be classified into four categories: local high point, insufficient well control, flow channel shielding and weak hydrodynamic. It is a systematic project to improve oil recovery in fracture-cavity carbonate reservoirs. In the stage of natural depletion, production should be well regulated to prevent bottom water channeling. In the early stage of waterflooding, injection-production relationship should be constructed according to reservoir type, connectivity and spatial location to enhance control and producing degree of waterflooding and minimize remaining oil. In the middle and late stage, according to the main controlling factors and distribution characteristics of remaining oil after water flooding, remaining oil should be tapped precisely by making use of gravity differentiation and capillary force imbibition, enhancing well control, disturbing the flow field and so on. Meanwhile, backup technologies of reservoir stimulation, new injection media, intelligent optimization etc. should be developed, smooth shift from water injection to gas injection should be ensured to maximize oil recovery.

Numerical simulation of pore-scale flow in chemical flooding process

Chemical flooding is one of the effective technologies to increase oil recovery of petroleum reservoirs after water flooding.Above the scale of representative elementary volume(REV), phenomenological modeling and numerical simulations of chemical flooding have been reported in literatures,but the studies alike are rarely conducted at the pore-scale,at which the effects of physicochemical hydrodynamics are hardly resolved either by experimental observations or by traditional continuum-based simulations.In this paper,dissipative particle dynamics(DPD),one of mesoscopic fluid particle methods,is introduced to simulate the pore-scale flow in chemical flooding processes.The theoretical background,mathematical formulation and numerical approach of DPD are presented.The plane Poiseuille flow is used to illustrate the accuracy of the DPD simulation,and then the processes of polymer flooding through an oil-wet throat and a water-wet throat are studies, respectively.The selected parameters of those simulations are given in details.These preliminary results show the potential of this novel method for modeling the physicochemical hydrodynamics at the pore scale in the area of chemical enhanced oil recovery.

Underwater Sea Cucumber Target Detection Based on Edge-Enhanced Scaling YOLOv4

Sea cucumber detection is widely recognized as the key to automatic culture.The underwater light environment is complex and easily obscured by mud,sand,reefs,and other underwater organisms.To date,research on sea cucumber detection has mostly concentrated on the distinction between prospective objects and the background.However,the key to proper distinction is the effective extraction of sea cucumber feature information.In this study,the edge-enhanced scaling You Only Look Once-v4(YOLOv4)(ESYv4)was proposed for sea cucumber detection.By emphasizing the target features in a way that reduced the impact of different hues and brightness values underwater on the misjudgment of sea cucumbers,a bidirectional cascade network(BDCN)was used to extract the overall edge greyscale image in the image and add up the original RGB image as the detected input.Meanwhile,the YOLOv4 model for backbone detection is scaled,and the number of parameters is reduced to 48%of the original number of parameters.Validation results of 783images indicated that the detection precision of positive sea cucumber samples reached 0.941.This improvement reflects that the algorithm is more effective to improve the edge feature information of the target.It thus contributes to the automatic multi-objective detection of underwater sea cucumbers.

基于CycleGAN-IA方法和M-ConvNext网络的苹果叶片病害图像识别

针对苹果叶片病害图像识别存在数据集获取困难、样本不足、识别准确率低等问题,提出基于多尺度特征提取的病害识别网络(Multi-scale feature extraction ConvNext,M-ConvNext)模型。采用一种结合改进的循环一致性生成对抗网络与仿射变换的数据增强方法(Improved CycleGAN and affine transformation,CycleGAN-IA),首先,使用较小感受野的卷积核和残差注意力模块优化CycleGAN网络结构,使用二值交叉熵损失函数代替CycleGAN网络的均方差损失函数,以此生成高质量样本图像,提高样本特征复杂度;然后,对生成图像进行仿射变换,提高数据样本的空间复杂度,该方法解决了数据样本不足的问题,用于辅助后续的病害识别模型。其次,构建M-ConvNext网络,该网络设计G-RFB模块获取并融合各个尺度的特征信息,GELU激活函数增强网络的特征表达能力,提高苹果叶片病害图像识别准确率。最后,实验结果表明,CycleGAN-IA数据增强方法可以对数据集起到良好的扩充作用,在常用网络上验证,增强后的数据集可以有效提高苹果叶片病害图像识别准确率;通过消融实验可得,M-ConvNex识别准确率可达99.18%,较原ConvNext网络准确率提高0.41个百分点,较ResNet50、MobileNetV3和EfficientNetV2网络分别提高3.78、7.35、4.07个百分点,为后续农作物病害识别提供了新思路。

Polarimetric entropy of the ocean surface with a two-scale scattering model

The relationships among an ocean wave spectrum,a fully polarimetric coherence matrix,and radar parameters are deduced with an electromagnetic wave theory.Furthermore,the relationship between the polarimetric entropy and ocean wave spectrum is established based on the definition of entropy and a twoscale scattering model of the ocean surface.It is the first time that the polarimetric entropy of the ocean surface is presented in theory.Meanwhile,the relationships among the fully polarimetric entropy and the parameters related to radar and ocean are discussed.The study is the basis of further monitoring targets on the ocean surface and deriving oceanic information with the entropy from the ocean surface.The contrast enhancement between human-made targets and the ocean surface with the entropy is presented with quad-pol airborne synthetic aperture radar（AIRSAR） data.

基于多尺度残差注意力网络的水下图像增强

针对水下图像由水的散射、吸收引起的色偏、色弱、信息丢失问题,提出了一种基于多尺度残差注意力网络的水下图像增强算法。该网络引入了改进的UNet3+-Avg结构与注意力机制,设计出多尺度密集特征提取模块与残差注意力恢复模块,以及由Charbonnier损失和边缘损失相结合的联合损失函数,使该网络得以学习到多个尺度的丰富特征,在改善图像色彩的同时又可保留大量的物体边缘信息。增强后图像的平均峰值信噪比(PSNR)达到23.63 dB、结构相似度(SSIM)达到0.93。与其他水下图像增强网络的对比实验结果表明,由该网络所增强的图像在主观感受与客观评价上都取得了显著的效果。

改进YOLOX的遥感图像目标检测算法

针对遥感图像目标检测算法复杂背景下目标检测精度低、小目标特征丢失的问题,提出一种改进YOLOX的遥感图像目标检测算法MYOLOX(modified YOLOX)。该算法在主干网络引入残差金字塔卷积模块(residual pyramid convolution module,RPCM)增强浅层特征图中的空间位置等细节信息,缓解下采样过程中的特征丢失。引入增强跨阶段局部块(improved cross stage partial block,ICSP)提取丰富的上下文信息并抑制噪声干扰,减少复杂背景及噪声干扰带来误检。将改进算法应用于使用DIOR数据集对NWPU VHR-10数据集扩充后数据集和SSDD数据集,MYOLOX算法检测平均精度均值(mean average precision,mAP)分别达到了80.8%和94.4%,较原算法提升了4.1和4.5个百分点。实验结果证明,改进后的算法能够明显提高遥感图像目标检测精度。

金字塔渐进融合低照度图像增强网络

针对现有低照度图像增强网络对不同尺度特征信息存在感知与表达能力不足的问题,提出金字塔渐进融合低照度图像增强网络模型。网络对图像进行多次下采样操作以组成特征金字塔,通过在特征金字塔的三个不同分支上加入跳跃连接,将不同尺度的特征图进行相互融合。通过精细恢复模块进一步提取精炼信息,将特征图恢复到正常的光照图像。结果表明,该网络模型不但能有效地提升低照度图像的整体亮度,而且能很好地保持图像中的细节信息和清晰的物体边缘轮廓,同时能够有效地抑制图像中的暗部噪声,使增强后的图像整体画面真实自然。

基于特征增强及多层次融合的火灾火焰检测

为提升火灾火焰识别检测方法性能,将传统图像处理与神经网络结合,提出1种基于特征增强及多层次融合的轻量级火灾火焰检测模型。模型利用多种色彩空间转换算法增强火焰特征信息,并设计双阶段多层次特征提取融合结构,配合空间注意力机制对火焰信息由粗到精进行提取;同时,针对火灾火焰特点,引入由浅到深逐步融合的自适应多尺度融合结构,提升对不同阶段火灾目标的检测精度。研究结果表明:本文模型可有效提升火灾火焰的检测效果,且具有更高的稳定性和鲁棒性,可准确高效地实现火灾火焰检测。研究结果可为现有火灾检测设备提供更准确的识别结果,从而更好地预防火灾事故发生。

基于多分支空谱特征增强的高光谱图像分类

为了解决高光谱图像自身及分类过程中噪声干扰大、空间-光谱特征信息提取不足以及有限样本下分类性能不佳等问题,提出一种基于多分支空谱特征增强的高光谱图像分类模型SSFE-MBACNN。首先,利用多分支特征提取模块分别提取浅层空谱特征和深层空间特征信息,并引入注意力机制抑制噪声干扰。其次,设计一种改进多尺度空谱特征提取融合模块及结合双池化和空洞卷积的空间特征增强模块实现空谱特征增强,减少模型参数量和提高分类性能。最后,用全局平均池化层代替全连接层,进一步降低参数量,缓解模型过拟合问题。实验结果表明,在Indian Pines(10%训练样本)、Pavia University (5%训练样本)和Salinas(1%训练样本)数据集分别取得了0.990 7、0.997 5和0.994 7的总体分类精度。SSFE-MBACNN不仅能充分利用空谱特征信息,而且在有限样本下也取得了优秀的分类性能,明显高于其他对比方法。

采用级联策略融合边界特征的多尺度息肉分割网络

结直肠息肉分割能有效辅助医生筛查大肠腺瘤,但息肉分割存在噪声较多、边界区分度不够等问题。针对以上问题,本文设计了一种采用级联策略融合边界特征的多尺度息肉分割网络。首先,本文提出了一种改进的通道分组空间增强模块,以增强骨干网络提取的图像特征,从而提高通道和空间位置的相关性。其次,考虑到边界区分度不够,设计了一个级联特征融合网络,以更好地保留边界信息并提高边界区分度,从而提高分割精度。最后,引入了一种双分支混合上采样模块来获取更多的特征细节信息,以实现特征的互补以及捕获更完整有效的特征。在CVC-ClinicDB和Kvasir数据集上进行测试,本文算法的平均Dice系数分别为0.944,0.920,平均交并比分别为0.900,0.869;而M2SNet算法的平均Dice系数分别为0.922,0.912,平均交并比分别为0.880,0.861。在ETIS-LaribPolypDB,CVC-300和CVC-ColonDB数据集上进行测试,本文算法的平均Dice系数分别为0.776,0.915,0.782;而M2SNet算法的平均Dice系数分别为0.749,0.903,0.758。实验结果表明本文算法的分割精度较高,泛化能力较强。