Salient region detection using background priors

In this paper, we introduce a novel salient region detection algorithm by using background priors. Because of the fact that superpixel is perceptually more meaningful than pixel, and which can reduce the complexity of image processing, we use the superpixel algorithm to reprocess original images. In addition, we hold the point that the colors in the image boundary could mainly represent all background colors, hence we compute the color contrast between the intern colors and the boundary colors. Since the nearer the patches are close to center, the more they affect other patches, we propose a new distribution-based model. Finally, experimental results demonstrate that the proposed method outperforms the state-of-the-art approaches.     

Evaluating nonlinear variability of mental fatigue behavioral indices during long‐term attentive task

This study investigates the behavioral indices of attention. A simple repetitive attentive task that resulted in mental fatigue was used consecutively in four trials. In the first step, reaction time and error responses were recorded to evaluate differences among trials. During the task, subjects showed different responses to stimulations. In the second part, to recognize the strategies, multiple clustering methods such as k‐means and fuzzy c‐means were performed in which behavioral indices and nonlinear features were used. In the last section, mental behavior was identified as a result of the chaotic properties of variations in reaction time. Therefore, the Lyapunov exponent of reaction times was evaluated. Results revealed that behavioral indices could distinguish attention from the occurrence of mental fatigue in trials. In addition, the three strategies used by subjects during the test protocol were assessed. Finally, variation of indices extracted from nonlinear analysis, that is, decrease in degree of chaotic behavior determined the transition from attention to mental fatigue. © 2012 Wiley Periodicals, Inc. Complexity, 2012     

融合图卷积神经网络和注意力机制的PM2.5小时浓度多步预测

PM_2.5小时浓度多为单步预测。为实现PM_2.5小时浓度的多步预测,基于“编码器-解码器”的序列-序列预测（Seq2Seq）模型,集合图卷积神经网络提取非欧式空间数据特征的能力以及注意力机制自适应关注特征的能力,提出了融合图卷积神经网络和注意力机制的PM_2.5小时浓度多步预测（GCN_Attention_Seq2Seq）模型。并与Seq2Seq模型和使用了图卷积神经网络、未使用注意力机制的GCN_Seq2Seq模型进行了对照,以2015—2016年北京市22个空气质量监测站点的空气质量数据为样本进行实例验证,结果表明,Seq2Seq模型和图卷积神经网络（GCN）可对PM_2.5小时浓度数据的时空依赖进行有效建模,注意力机制有助于减缓多步预测中的预测精度衰减,提升PM_2.5小时浓度多步预测的精度。GCN_Attention_Seq2Seq模型可有效应用于多种长度的PM_2.5浓度预测窗口。     

Learning Visible Thermal Person Re-Identification via Spatial Dependence and Dual-Constraint Loss

Visible thermal person re-identification (VT Re-ID) is the task of matching pedestrian images collected by thermal and visible light cameras. The two main challenges presented by VT Re-ID are the intra-class variation between pedestrian images and the cross-modality difference between visible and thermal images. Existing works have principally focused on local representation through cross-modality feature distribution, but ignore the internal connection of the local features of pedestrian body parts. Therefore, this paper proposes a dual-path attention network model to establish the spatial dependency relationship between the local features of the pedestrian feature map and to effectively enhance the feature extraction. Meanwhile, we propose cross-modality dual-constraint loss, which adds the center and boundary constraints for each class distribution in the embedding space to promote compactness within the class and enhance the separability between classes. Our experimental results show that our proposed approach has advantages over the state-of-the-art methods on the two public datasets SYSU-MM01 and RegDB. The result for the SYSU-MM01 is Rank-1/mAP 57.74%/54.35%, and the result for the RegDB is Rank-1/mAP 76.07%/69.43%.     

基于双阈值视觉注意模型的图像关注焦点检测

 图像关注焦点（FOA）检测是基于人眼视觉关注模型的图像感兴趣区提取的关键技术。为了更加精确、合理地搜索图像关注焦点,提出一种基于双阈值视觉关注模型的FOA检测算法。算法首先提取图像的亮度、颜色、方向和离散矩变换（DMT）特征,生成各个特征对应的特征图;然后将各特征图合并为一张包含多种特征的显著图;最后根据显著图的灰度直方图建立静态阈值与动态阈值,确定图像关注焦点的位置与数量。实验结果表明：新算法在图像关注焦点检测的准确性上较Itti模型有更为优秀的表现,更符合人眼视觉的关注习惯。     

基于Mel频谱值和深度学习网络的鸟声识别算法*

为了增强网络对鸟鸣声信号的特征学习能力并提高识别精度,提出一种基于深度残差收缩网络和扩张卷积的鸟声识别方法。首先,提取鸟鸣声信号的对数梅尔特征及其一阶和二阶差分系数组成logMel特征集作为网络模型的输入;其次,通过深度残差收缩网络自动学习噪声阈值,减少噪声干扰;然后,引入扩张卷积增大卷积核感受野并利用注意力机制使网络更关注关键帧特征;最后,通过双向长短时记忆网络从学到的局部特征中学习长期依赖关系。以百鸟数据birdsdata鸟声库中的19种中国常见鸟类作为实验对象,识别正确率可以达到96.58%,并对比模型在不同信噪比数据下的识别结果,结果表明该模型在噪声环境下的识别效果优于现有模型。     

改进的Faster-RCNN算法在聚乙烯管接头内部缺陷检测中的应用

超声相控阵技术是目前聚乙烯管道热熔接头内部缺陷检测的一种主流方法。提出了基于注意力机制的改进Faster-RCNN目标检测网络用于超声相控阵D扫图聚乙烯管接头内部缺陷检测。针对聚乙烯管道热熔接头内部超声相控阵D扫图小缺陷较多、特征信息容易丢失的问题,将残差网络(ResNet50)与特征金字塔网络(FPN)相结合作为骨干网络,并引入卷积注意力模块(CBAM)自适应细化特征。将SSD网络框架和Faster-RCNN网络框架用于模型训练和测试,使用VGG16、ResNet50、ResNet50+FPN、ACBM+ResNet50+FPN作为骨干网络依次对超声相控阵聚乙烯管道热熔对接接头内部缺陷样本进行训练对比。结果表明,改进的Faster-RCNN网络模型在聚乙烯管接头内部缺陷检测和分类方面有明显改进,对小缺陷的检测性能有了显著的提高。     

Heart rate variability analysis by chaotic global techniques in children with attention deficit hyperactivity disorder

Attention deficit hyperactivity disorder (ADHD) is characterized by decreased attention span, impulsiveness, and hyperactivity. Autonomic nervous system imbalance was previously described in this population. We aim to compare the autonomic function of children with ADHD and controls by analyzing heart rate variability (HRV). Children with ADHD (22 boys, mean age 9.964 years) and 28 controls (15 boys, mean age 9.857 years) rested in supine position with spontaneous breathing for 20 min. Heart rate was recorded beat by beat. HRV analysis was performed by use of chaotic global techniques. ADHD promoted an increase in the chaotic forward parameter. The algorithm which applied all three chaotic global parameters was only the second optimum statistically measured by Kruskal–Wallis (P < 0.0001) and low standard deviations. It was also highly influential by principal component analysis with almost all variation covered by the first two components. The third algorithm which lacked the (high spectral Detrended Fluctuation Analysis) parameter performed best statistically. However, we chose the algorithm which applied all three chaotic globals due to previous studies mentioned in the text—forward and inverse problems. Comparison of the autonomic function by analyzing HRV with chaotic global techniques suggests an increase in chaotic activity in children with ADHD in relation to the control group. © 2015 Wiley Periodicals, Inc. Complexity 21: 412–419, 2016     

Personal Interest Attention Graph Neural Networks for Session-Based Recommendation

Session-based recommendations aim to predict a user’s next click based on the user’s current and historical sessions, which can be applied to shopping websites and APPs. Existing session-based recommendation methods cannot accurately capture the complex transitions between items. In addition, some approaches compress sessions into a fixed representation vector without taking into account the user’s interest preferences at the current moment, thus limiting the accuracy of recommendations. Considering the diversity of items and users’ interests, a personalized interest attention graph neural network (PIA-GNN) is proposed for session-based recommendation. This approach utilizes personalized graph convolutional networks (PGNN) to capture complex transitions between items, invoking an interest-aware mechanism to activate users’ interest in different items adaptively. In addition, a self-attention layer is used to capture long-term dependencies between items when capturing users’ long-term preferences. In this paper, the cross-entropy loss is used as the objective function to train our model. We conduct rich experiments on two real datasets, and the results show that PIA-GNN outperforms existing personalized session-aware recommendation methods.     

视觉注意机制在图像增强中的应用研究

将视觉注意机制引入到直方图构造中,并在此基础上提出了一种新的基于灰度级信息量直方图的图像增强算法.该算法利用Itti视觉注意计算模型对图像的显著性进行分析,获得全局显著图;然后,将全局显著图划分为若干等大的子区域,求取各子区域的平均显著值,并做归一化处理,得到子区域的加权统计系数;再将各子区域的灰度级加权统计值相加,得到灰度级信息量直方图;最后,依据直方图均衡化的映射函数,调整灰度级的动态范围.实验结果表明,该算法明显优于经典的GHE算法和AHE算法,具有满意的视觉效果.