基于加权LoG算子的红外弱小目标检测方法研究

提出一种基于加权LoG算子及形态学运算的方法来提高远距离红外弱目标检测效率。通过不同尺度的加权LoG算子对图像进行运算,提取响应值最大的特征图。对图像进行形态学运算去噪,并进行Otsu二值分割聚类。输出目标点在图像上的位置坐标。实验结果表明:该方法与传统滤波方法相比,信噪比增益为36.9,杂波抑制因子为4.7,均比传统滤波方法要好。     

层次卷积滤波红外弱小目标检测方法

针对单帧复杂背景红外图像点目标检测算法存在复杂背景下处理效果不理想、处理时间长的问题,提出了一种层次卷积滤波检测算法。主要分为两个部分：第一,根据红外小目标特性,设计一种层次卷积滤波的算子,对图像进行滤波处理,实现图像中小目标的增效和背景抑制的效果;第二,采用基于最大值的自适应阈值方法,对图像进行二值化操作,过滤背景杂波,最终提取到待检测的目标。在大量不同背景红外图像中进行实验,论文算法在背景抑制因子和信噪比增益的性能量化结果上优于现有5种典型红外弱小目标检测算法的性能结果,且平均处理时间仅为高斯拉普拉斯（Laplacian of Gaussian,LoG）滤波算法的30.42%。通过实验对比,表明该层次卷积滤波算法可以有效解决在不同复杂背景下的红外图像中对小目标检测的问题。     

各向异性SUSAN滤波红外弱小目标检测

为了解决SUSAN滤波算子不能自适应调整滤波系数的问题,采用Geusebroek提出的各向异性高斯滤波器替代SUSAN滤波算子中的高斯滤波部分。由局部图像的方差和像素的邻域平滑度决定长短轴的方差,由该点的梯度方向决定滤波器的长轴方向,由局部图像的灰度值与均值差的一阶范数确定SUSAN滤波器的阈值,从而构造出各向异性SUSAN滤波器。将其用于红外弱小目标检测中,实验结果表明:各向异性SUSAN滤波器能够很好地保留图像中的边缘信息,使残差图像中弱小目标的信噪比增益和信杂比增益极大地提高,目标大小得到较好的保留,虚警率下降。     

多尺度形态算子融合图像滤波技术及滤波质量评价

针对舰载红外警戒系统的红外和电视图像,提出了一种新的海空背景下受强杂波、噪声污染的图像目标滤波算法和滤波效果的定量评价算子。算法采用多尺度的形态算子对输入的图像并行滤波,大尺度形态算子抑制图像噪声,小尺度形态算子提取目标边缘细节信息。处理后的图像进行基于树状小波帧变换的图像信息融合,融合图像可完备提取不同尺度滤波后的图像信息。针对目标检测跟踪的图像滤波算法的评价,提出了目标与背景的交叉分辨力评价算子及评价准则。仿真实验表明。该滤波算法要优于中值滤波、自适应滤波、小波变换滤波算法,滤波质量的定量评价算法是合理的、有效的。算法适用于舰载红外警戒系统。     

一种视频运动目标精确分割新算法

针对视频图像中单个运动目标的分割问题,提出了一种基于Kirsch边缘算子的视频运动目标分割算法,该算法将Kirsch算子检测到的边缘作为主分割信息,运动矢量场作为次要分割信息。首先利用双重尺度的运动矢量场进行累加和滤波处理来获得辅助分割信息;然后将Kirsch算子的模板分解为差值模板和公共模板以提高边缘的抗噪性;最后用自适应状态标记的方法将边缘信息和运动矢量信息相融合来准确地分割运动目标。实验结果表明该方法分割比较精确。     

红外弱小目标检测中的特征选择性滤波方法

针对红外图像弱小目标检测中的预处理问题,提出了一种特征选择性滤波方法(CSF),该方法创造性地将各向异性滤波应用到图像空域,通过分析红外图像中不同特征区域的特性,改进了系数分布函数,使CSF在消除背景、平滑边缘的同时,增强目标信号.最后给出了该方法的滤波特性分析和实验结果,并与其他滤波方法做了比较.实验结果表明该方法具有良好的背景抑制和目标增强性能,且结构简单,有利于实时实现.     

基于旋转SURF算子的图像配准新方法

针对SURF算法中快速Hessian矩阵行列式检测出的特征点的不连续现象,从而造成的旋转,模糊和光照变化适应性较差的不足,提出一种旋转SURF检测算子的图像配准新方法。该算法通过将SURF算法的积分图像盒子滤波模板逆时针旋转45度,引入一种可以检测角度旋转的滤波核提升检测算子对不同图像变换的匹配性能,保证新的检测算子与原算法较好的结合,同时利用改进的单纯形算法依据输入图像进行参数优化。仿真结果表明,该方法不仅保留了算法的速度优势,缩短了配准时间,而且在图像模糊变换,光照变换和JPEG压缩变换方面性能有明显的提升,此外对视角变换以及小尺度变换性能也有提高。     

基于自适应双层TDLMS滤波的红外小目标检测

提出了一种自适应滤波的红外小目标检测算法,该方法由背景去除与目标提取两层二维最小均方滤波器构成.分别引入背景模板与目标模板,结合二维最小均方滤波方法对原始红外图像进行处理,有效地剔除了背景噪声与杂波,同时尽可能地保留了目标信息.在算法实现过程中,提出了一种步长自动调整算法,通过图像的统计参数自适应调整步长的大小,可迭代出TDLMS滤波器的最优权值,再与权重模板相融合,有效增强了目标的检测效果.实验结果表明,该方法能够适应不同背景下的目标检测,有效提高了红外小目标的检测性能.     

采用视觉细胞响应模型的小目标复杂背景抑制

为了提高地面和云层等红外复杂背景下弱小目标的检测性能,提出了一种基于视觉细胞响应模型的红外弱小目标背景抑制新方法.首先利用简单细胞的感受野计算模型将原始图像采用Gabor函数卷积获得相同大小的两幅图像|然后采用设计的复杂细胞响应的非线性汇聚策略函数对获得的两幅图像进行融合处理,从而将红外图像中弱小目标和背景杂波分离,达到抑制背景的目的|最后采用自适应阈值分割技术得到目标点,实现了对红外弱小目标的检测跟踪.实验结果显示,与去局部均值和最大中值滤波两种滤波方法相比较,该方法能有效地检测出信杂比较低的弱小目标信号.     

一种抗干扰的弱小目标检测方法

针对低信噪比灰度图像中弱小目标的检测问题,提出了SUSAN特征检测原则与数学形态学方法结合的小目标检测方法.该方法首先采用SUSAN原则利用图像灰度级直接提取特征,利用形态学算子去除噪声、分离目标.给出了应用实例,实验结果表明,这种方法抗干扰能力强,能够快速、可靠地检测出小目标.     

Partial Trace Method for Deriving Density Operators of Light Field

With the help of technique of integration within an ordered product of operators we can fully display the partial trace method to deduce some new density operators of light field. The general form of single-mode density operator is derived in this way from a normalized two-mode generalized squeezed pure state density operator.     

基于Top-hat变换的OSAHS图像边缘检测算法

提出了一种基于多方向、多尺度Top-hat变换的图像边缘检测方法,应用于阻塞性睡眠呼吸暂停低通气综合症(Obstructive Sleep Apnea Hypopnea Syndrome, OSAHS)早期病理图像的边缘检测及诊断。首先,构造不同方向、不同尺度的Top-hat算子增强图像的对比度,利用形态学梯度进行边缘检测,然后把各个算子检测到的图像边缘按照一定的权重进行组合,得到理想的边缘,以便准确地获得病理图像的相关参数,进而实现医学电子诊断。本文以口腔图像、咽喉声带处图像、鼻道内部图像的处理为例,这三组图像的处理结果表明,与传统的边缘算子相比较,该方法能使图像的边缘信息更完整、更准确,图像的边缘闭合度可达到97.67%。     

Infrared dim small target enhancement using toggle contrast operator

A new infrared dim small target enhancement algorithm based on toggle contrast operator is proposed. Toggle contrast operator is modified and used to construct operators using the image features derived from dilation and erosion operators. Then, based on the constructed operators, the operators which could be used to estimate the clutter background of the original infrared dim small target image are proposed using the same strategy as the definition of opening. Finally, the infrared dim small target is well enhanced through subtracting the estimated background from the original image. Experimental results on infrared images with different types of targets verified that the proposed method could effectively enhance infrared dim small target, which would be very useful for infrared dim small target detection and tracking.     

Real-time application of critical dimension measurement of TFT-LCD pattern using a newly proposed 2D image-processing algorithm

A critical dimension measurement system for TFT-LCD patterns has been implemented in this study. To improve the measurement accuracy, an imaging auto-focus algorithm, fast pattern-matching algorithm, and precise edge detection algorithm with subpixel accuracy have been developed and implemented in the system.The optimum focusing position can be calculated using the image focus estimator. The two-step auto-focusing technique has been newly proposed for various LCD patterns, and various focus estimators have been compared to select a stable and accurate one.Fast pattern matching and subpixel edge detection have been developed for measurement. The new approach, called NEMC, is based on edge detection for the selection of influential points; in this approach, points having a strong edge magnitude are only used in the matching procedure. To accelerate pattern matching, point correlation and an image pyramid structure are combined.Edge detection is the most important technique in a vision inspection system. A two-stage edge detection algorithm has been introduced. In the first stage, a first order derivative operator such as the Sobel operator is used to place the edge points and to find the edge directions using a least-square estimation method with pixel accuracy. In the second stage, an eight-connected neighborhood of the estimated edge points is convolved with the LoG (Laplacian of Gaussian) operator, and the LoG-filtered image can be modeled as a continuous function using the facet model. The measurement results of the various patterns are finally presented.The developed system has been successfully used in the TFT-LCD manufacturing industry, and repeatability of less than 30 nm (3σ) can be obtained with a very fast inspection time.     

融合区域生长与霍夫变换的内窥图像分割算法

为准确地划分出实际内窥图像的有效检测区域,依据此类图像的具体特点提出一种综合区域生长和霍夫变换的分割算法。利用区域生长大致分割出感兴趣区域,可能会存在漏检边缘或虚假边缘,通过二值形态学处理对图像进行平滑滤波和去噪,采用Canny算子在抑制噪声的同时进行边缘检测,应用霍夫变换检测圆的算法确定图像内有效区域的位置。通过对90组实际内窥图像在Visual C++ 6.0上进行仿真,实验结果表明:有88组内窥图像能够精确地分割强光干扰且划分出有效检测区域;仅有2组图像分割出的强光干扰及划分出的有效检测区域不够准确。     

Image denoising using least squares wavelet support vector machines

We propose a new method for image denoising combining wavelet transform and support vector machines (SVMs). A new image filter operator based on the least squares wavelet support vector machines (LSWSVMs) is presented. Noisy image can be denoised through this filter operator and wavelet thresholding technique. Experimental results show that the proposed method is better than the existing SVM regression with the Gaussian radial basis function (RBF) and polynomial RBF. Meanwhile, it can achieve better performance than other traditional methods such as the average filter and median filter.     

An improved infrared dim and small target detection algorithm based on the contrast mechanism of human visual system

To achieve higher detection rate and lower false alarm rate in dim and small target detection, this paper proposed an improved algorithm based on the contrast mechanism of human visual system (HVS) for infrared small target detection in an image with complicated background. According to the contrast mechanism of HVS, Laplacian of Gaussian (LoG) filter is exploited to deal with the input image, which can not only suppress the background noise and clutter but also enhances the target intensity significantly. As a result it increases the contrast ratio between target and background. To further eliminate residual clutter, we process the filtered image with morphological method in all directions. True target is finally obtained by applying local thresholding segmentation to the pre-processed image. Experimental results demonstrate its superior and reliable detection performance by high detection rate and low false alarm rate.     

空间信息自适应融合的高光谱图像分类方法

针对单一的滤波器提取高光谱图像空间纹理信息时不能获得完整的图像特征的不足,提出一种结合双边滤波和域转换标准卷积滤波的高光谱图像分类算法.该方法采用空间信息自适应融合的分类寻优,先对高光谱波段进行抽样分组,再用双边滤波和域转换标准卷积滤波对分组后的波段进行滤波,两种空间信息进行线性融合后交由支持向量机完成分类.实验表明,相比使用光谱信息、高光谱降维、空谱结合的支持向量机分类方法和边缘保持滤波以及递归滤波的方法,本文所提算法对高光谱图像的分类精度有较大提高,在训练样本仅为5%和3%的情况下,对印第安农林和帕维亚大学图像的总体分类精度分别达到了96.95%和97.89%,比其他算法高出213个百分点,验证了该方法在高光谱图像分类的有效性.