图象处理在物体表面检测中的应用

本文介绍了利用计算机图象处理技术对物体表面检测的原理及方法，该文中还提出了各种算法的软件实现。     

模数变换器动态测试系统中数据获取接口的研制

本文介绍了模数变换器动态系统中数据获取接口的研制。该接口设计的难点是大容量和高速度。文中介绍了分组交替存取技术和加一运算器的设计方法，该接口设计方法对设计其他高速数据采集和实时信息处理系统也有参考意义。     

Three-dimensional metrology of surface extracted from a cloud of measured points using a new point-to-surface assignment method: An application to PCB-mounted solder pastes

This work presents a method of point-to-surface assignment for three-dimensional metrology of solder paste bricks on printed circuit board (PCB). A bounding box enclosing the brick tightly on all sides is introduced to avoid incorrect point-to-surface assignment. The shape of bounding box for solder paste brick is variable according to geometry of measured points. The surface geometry of the bounding box is obtained by using five peaks selected from the histogram of normalized gradient vectors for measured points. By using the bounding box enclosing the solder paste brick, the task of point-to-surface assignment is successfully executed. Subsequently, the geometrical features are obtained via surface fitting.     

允许姿态变化的快速人脸特征检测

本文提出一种在允许姿态显著变化的人脸图象上快速检测眼睛和嘴角的方法。该方法仅仅合理地假设人脸图象要以通过值分割将头发和眼睛与脸部明显分开。首先采用Ｈｏｕｇｈ变换并利用有脸的相似性和对称性约束确定两个瞳孔的位置，然后根据人体测量关系初略估计嘴部区域，最事利用积分投影方法得到嘴角的精确位置。     

Deep network based stereoscopic image quality assessment via binocular summing and differencing

With the development of deep networks in dealing with various visual tasks, the deep network based on binocular vision is expected to tackle the issue of stereoscopic image quality assessment. Here, we present a stereoscopic image quality assessment method using the deep network with four channels together, which takes the left view, right view, binocular summing view, and binocular differencing view as the inputs of the network. The visual features are enhanced through the concatenation in a weighted way, so that the binocular vision can be adequately included in the binocular addition and subtraction information. Compared with the state-of-the-art metrics, the proposed method exhibits relatively high performances on four benchmark databases.     

基于密集卷积神经网络的遥感飞机识别

传统的飞机识别方法受模糊、遮挡、噪声以及光照等多种因素的干扰时会降低识别率,且卷积神经网络主要依赖局部特征,却丢失了轮廓特征等重要的全局结构化特征,从而导致算法对于受干扰飞机图像识别效果不佳。因此,基于密集卷积神经网络提出一种结合局部与全局特征的联合监督识别方法,以密集卷积神经网络为基础得到图像特征,通过结合局部特征（卷积神经网络特征）与全局特征（方向梯度直方图特征）进行分类,分类器目标函数使用softmax损失和中心损失联合监督方法。实验结果表明,局部特征与全局特征的结合使算法更加智能化,且损失函数联合监督方法能够实现图像深层特征的类内聚合、类间分散,该算法能有效解决卷积神经网络对受到多种干扰的遥感图像识别率低的问题。     

基于像素值排序的可逆信息隐藏算法

针对嵌入秘密后灰度图失真明显的问题,提出一种基于像素值排序（PVO）的可逆信息隐藏算法。首先,将像素分成灰、白两层,选择灰层的像素作为目标像素,对目标像素十字交叉位置上的4个白色像素进行排序;然后根据排序结果计算两端两个像素的均值和中间两个像素的均值,利用可逆约束实现像素的动态预测;最后,根据预测结果构造预测误差直方图（PEH）,使用环形复杂度实现秘密数据的自适应嵌入,并用同样的方法处理白色层像素。利用USC-SIPI标准图像库中6幅图像进行仿真实验,当嵌入容量（EC）为10000 b,平均峰值信噪比（PSNR）为61.89 dB时,该算法能有效减小携密图像的失真。     

Remote Sensing Estimation of Forest Coverage based on Histogram and Pixel Decomposition Model

Forest coverage is the percentage of forest cover for hooking out the forest stand boundary,and quantitative coverage information can be used for describing the temporal and spatial variability of vegetation in the horizontal scale.Pixel decomposition model has been widely used in remote sensing estimation of vegetation coverage.However,there are still many problems.For example,it is difficult to find a pure spectrum of tree canopy coverage to estimate the crown coverage with high accuracy in forest vegetation applications.In this paper,combining land use and soil type data,it is proposed to determine the endmember parameters of the different vegetation-soil types by using the histogram method based on pixel decomposition model for estimating the regional scale forest coverage in the Three Gorges.And the results were verified using the 161 samples of field data in the Three Gorges Reservoir area,R² was found to be 0.742 4~0.853 6,the estimated results are satisfactory.This method will provide a reference for remote sensing estimation of high-resolution forest coverage at regional scale.     

基于HOG与多实例在线学习的目标跟踪算法

为实现在局部遮挡、光线变化等复杂背景下的目标跟踪,提出一种基于梯度方向直方图(HOG)与多实例在线学习的目标跟踪算法.利用已标定目标图像的HOG特征空间,结合局部二值模式(LBP)描述方法获取特征向量,构建初始随机蕨检测算子,采用随机多尺度采样方法跟踪每一帧的目标位置和尺寸,并基于多实例在线学习框架,通过检测到的目标样本以及附近的背景样本在线更新检测算子.将该算法与OnlineB oostingTracker,MILTracker等在线学习目标跟踪算法在多个标准视频序列中进行比较,实验结果表明,该算法在局部遮挡和光照变化的环境下具有较好的跟踪稳定性,但在抗目标旋转方面有待优化.     

基于DRPKP算法的文本去重研究与应用

SimHash算法是目前主流的文本去重算法,但它对于特定行业的文本数据在主题方面的天然相似性特点并没有特殊的考虑.基于多年在金融证券行业信息管理和数据整合的经验,本文分析目前文本去重方法存在的问题,特别针对SimHash算法在特定行业文本去重中的不足,创新地提出一种基于段落主题的文本去重方法（简称DRPKP算法）,通过对去重准确率、覆盖率和去重时间3个指标进行对比测试,DRPKP算法比SimHash算法准确率可提高24.5％、覆盖率可提高16.34％,且去重时间更短.