采用最小误差阈值分割算法的基因芯片图像分析

为了能够较好地处理芯片图像,尽可能准确地提取出描述基因样点的数据信息,采用了最小误差阈值的分割算法.该方法在假设目标和背景的分布服从混合正态分布的前提下,设定了最小误差分类目标函数,通过求得使目标函数值最小的最佳分割阈值,实现基因样点和背景图像的分割.针对分割出来的基因样点图像提取特征数据,最后对这些数据进行聚类分析,进而对实验样点进行分类.在实验中应用该方法分析了2组基因芯片图像,基因样点的分类效果较好,验证了该基因芯片分析方法的可行性.     

微阵列样点定位与信息提取

提出了一种自动划格方法用于定位cDNA微阵列图像中的样点,这种方法对于解决大量微阵列图像数据处理优势明显。结合局部阈值和对数阈值的处理方法对图像进行样点提取,为信息提取提供更高的精度。将自动划格方法用于定位微阵列图像,有效地避免了人工参与带来的干扰,提高了微阵列图像处理速度。     

基于图像处理的微小电子元器件自动计数算法

根据微小电子元器件的形状和正反两面的纹理特征,提出一种基于图像处理的微小电子元器件自动计数算法.该算法采用层次特征提取及数学形态学的图像分割算法,对图像中的元器件正面和反面,分别进行图像处理.根据元器件反面亮度高及纹理简单的特点,采用阈值化方法对其进行完全分割.将正面朝上的元器件从图像中分割出来,并采用数学形态学方法分离粘连的正面元器件.从而对粘连在一起的微小电子元器件实现了完全分割.对元器件进行精确分割后,即可采用连通区域标记的方法对所有分离的元器件进行计数.该算法实现了微小电子元器件实时自动计数,计数精度达到百分之百.     

改进的机场停泊飞行器目标分割算法

考虑到飞行器目标在整幅图像中所占的比例往往较小,且图像背景复杂,本文提出了一种基于机场区域提取的飞行器目标分割算法.该算法首先利用Hough变换检测直线的特性,定位机场跑道和停机坪的位置,并结合教学形态学等图像处理技术去除了非机场区域;在提取机场区域后,再选择适当的阈值对图像进行分割,最后经过形态学去噪、小区域去除等步骤分割出飞行器目标.实验结果表明,该算法改进了以往机场区域提取算法保留候机楼等附属部分以及提取结果中存在机场区域以外区域的缺点,较好地实现了机场停泊飞行器目标的分割,为下一步准确识别飞行器类型奠定了基础.     

基于数学形态学的医学图像分割方法研究

文章介绍了医学图像处理中的医学图像分割的有关概念和数学形态学进行图像分析的基本步骤,重点论述了几种医学图像分割方法和基于数学形态学的分水岭分割算法,并给出了该算法的优点。     

数学形态学在图像分割中的应用研究

本文以数学形态学基本运算的介绍为基础,阐述了图像处理中常见的一些数学形态学方面的应用,其中有形态学图像重建、形态学图像滤波和形态学图像梯度。最后介绍了一种基于形态学的图像分割算法—分水岭算法,并以基于标记的分水岭分割算法为例,研究了形态学运算在图像分割前的预处理步骤和图像分割中帮助提取标记的作用,由此证明了数学形态学运算是图像处理领域中的一种有效方法和手段。     

基于机器视觉的蟋蟀体态测量

设计一套基于机器视觉的蟋蟀体态测量装置,初步研究用图像处理和图像测量的方法实现对蟋蟀躯干长宽的测量.该装置构建一个PC-Based的机器视觉系统,利用VisuaI Basic开发工具编写图像处理算法和图像测量算法的程序.应用惯性主轴的方法实现蟋蟀图片位置的归一化,应用数学形态学分割算法实现了蟋蟀躯干与足和触角的分割,将分割后的躯干进行边缘提取,从而进行蟋蟀体态的测量.     

基于形态学梯度的分水岭彩色图像分割

图像分割是从图像处理到图像分析的关键步骤之一。在改进了基于地形学距离的分水岭算法的基础上,提出了一种结合图像信息熵、形态学梯度与区域合并的图像分割方法。该算法首先利用信息熵在RGB颜色空间中对彩色图像求其形态学梯度,然后对彩色梯度图进行分水岭分割,最后对分水岭产生的过分割现象进行区域合并。通过Matlab对图像进行实验,结果证明该算法不仅能够减少分水岭算法的过分割现象,而且还提高了图像分割的精确性,同时在图像分割时具有很好的鲁棒性和适应性。     

基于小波变换和形态学分水岭的血细胞图像分割

医学图像处理提取细胞中使用分水岭方法时,容易产生过分割现象且对噪声的干扰极为敏感,为了解决此缺点,提出一种基于小波变换和形态学分水岭的细胞图像分割新方法。首先采用小波变换多分辨率分析对图像进行分解,选取合适的小波基和改进去噪阈值函数对图像进行小波去噪,然后对去噪后小波重构的细胞图像应用数学形态学距离变换、灰度重建等技术产生的区域标记进行分水岭变换,最终得到分割结果。实验结果表明,该算法能稳定、准确地提取细胞和实现粘连细胞的自动分割,同时具有很好的鲁棒性和普适性。     

基于自动随机游走的涎腺肿瘤超声图像分割方法

针对涎腺超声图像斑点噪声强、对比度低和边界弱的特点,提出了一种结合形态学检测的自动随机游走分割方法.该方法首先利用形态学操作获得目标的初始轮廓,然后提取目标区域和背景区域骨架结构的有效标记点作为随机游走算法的种子点,最后利用种子点对预滤波后的肿瘤图像实现随机游走分割.实验选取大量临床采集的涎腺肿瘤超声图像进行测试,结果表明该方法计算复杂度低,解决了传统随机游走模型初始种子点的人工干预问题,有效实现了涎腺肿瘤的自动分割.     

Robust adaptive spot segmentation of DNA microarray images

The rapid advancement of DNA chip (microarray) technology has revolutionalized genetic research in bioscience. However, the enormous amount of data produced from a microarray image makes automatic computer analysis indispensable. An important first step in analyzing microarray image is the accurate determination of the DNA spots in the image. We report here a novel spot segmentation method for DNA microarray images. The algorithm makes use of adaptive thresholding and statistical intensity modeling to: (i) generate the grid structure automatically, where each subregion in the grid contains only one spot, and (ii) to segment the spot, if any, within each subregion. The algorithm is fully automatic, robust, and can aid in the high throughput computer analysis of microarray data.     

DNA微阵列图象信息的自动提取

微阵列图象分析是分子生物学中DNA微阵列杂交实验数据测定过程的一个重要部分。其目的是将微阵列图象中大量象素灰度信息简化为微阵列靶点的信号值。该文介绍了一种对DNA微阵列图象信息进行自动提取的方法,使用此方法完全不需要人为操作,可避免操作者主观性对实验的影响,提高数据提取的效率和可重复性。其关键步骤包括:图象滤波,图象灰度信息提取,微阵列间距确定,靶点定位和靶点信号值计算。实验结果表明使用此算法提取DNA微阵列图象信息具有重复性好、效率高和分析准确的特点。     

一种基于聚类和统计分析DNA基因芯片图像处理算法

DNA基因芯片可以同时监控成千上万个基因的表达信息。图像分析是基因芯片试验中一个重要的环节,直接影响到其后续的处理、分析和研究,比如鉴别预测具有不同表达信息的基因功能。基因芯片图像分析包括三个步骤:图像网格化,图像分割以及信息抽取。该文主要研究分割和信息抽取问题。首先基于K-Means聚类技术提出了一种新的分割方法;其次基于统计分析文章建议了一种新的背景和前景分割校正方法用于更准确的信息抽取。新方法的优点是对于基因芯片中spot图像没有任何形状限制。实际图像分析结果与目前最流行的基因芯片图像分析软件GenePix对比研究表明该文算法是精确有效的。     

A noise-resistant algorithm for grid finding in microarray image analysis

We present an algorithm for automatic spot localization for microarray images with rectangular spot and block packing. As an input, the algorithm requires only the common array design parameters: number of block rows and columns and number of spot rows and columns within each block. It proved to be robust with respect to different types of contamination and can tolerate a high percentage of the missing spots. The validity of the developed algorithm has been tested and confirmed using a large set of images of various designs from different microarray platforms. Comparison with academic and commercial packages has shown that for uncontaminated images our algorithm performs similarly, whereas for certain problematic images it outperforms the other packages.Patent pending.     

Robust segmentation and analysis of DNA microarray spots using an adaptative split and merge algorithm

Microarray images push to their limits classical analysis methods, since gene spots are often poorly contrasted, ill defined and of irregular shapes. These characteristics hinder a robust quantification of corresponding values for red and green intensities as well as their R/G ratio. New approaches are thus needed to ensure accurate data extraction from these images. Herein we present an automatic non-supervised algorithm for a fast and accurate spot data extraction from DNA microarrays. The method is based on a split and merge algorithm, relying on a Delaunay triangulation process, allowing an incremental partition of the image into homogeneous polygons. Geometric properties of triangles as well as homogeneity criteria are defined according to the specificities of microarray image signals. The method is first assessed on simulated data, and then compared with GenePix and Jaguar Softwares. Results in segmentation and quantification are superior to those obtained from a number of standard techniques for spot extraction.     

Segmentation of Microarray Images by Mathematical Morphology

DNA chips (i.e., microarrays) biotechnology is a hybridization (i.e., matching of pairs of DNA)-based process that makes possible to quantify the relative abundance of mRNA of two distinct samples by analyzing their fluorescence signals. This technique requires robotic placement (i.e., spotting) of thousands of cDNAs (i.e., complementary DNA) in an array format on glass microscope slides. The spotted cDNAs are the hybridization targets for the mRNA samples. The two different samples of mRNA, usually labeled with Cy3 and Cy5 fluorochromes, are cohybridized onto each spotted gene. After hybridization, one digital image is acquired for each fluorochrome wavelength. Then, it is necessary to recognize each gene by its position in the array and to estimate its signal (i.e., hybridization information). For that, it is necessary to segment the image in three classes of objects: subarrays (i.e., set of grouped spots), spot box (i.e., the rectangular neighborhood that contains a spot) and spot (i.e., region of the image where there exists signal). In this paper, we present a technique based on mathematical morphology that performs this segmentation. In the website http://www.vision.ime.usp.br/demos/microarray/detailed experimental results are presented.     

On denoising and compression of DNA microarray images

We address the problems of noise and huge data sizes in microarray images. First, we propose a mixture model for describing the statistical and structural properties of microarray images. Then, based on the microarray image model, we present methods for denoising and for compressing microarray images. The denoising method is based on a variant of the translation-invariant wavelet transform. The compression method introduces the notion of approximate contexts (rather than traditional exact contexts) in modeling the symbol probabilities in a microarray image. This inexact context modeling approach is important in dealing with the noisy nature of microarray images. Using the proposed denoising and compression methods, we describe a near-lossless compression scheme suitable for microarray images. Results on both denoising and compression are included, which show the performance of the proposed methods. Further experiments using the results of the proposed near-lossless compression scheme in gene clustering using cell-cycle microarray data for S. cerevisiae showed a general improvement in the clustering performance, when compared with using the original data. This provides an indirect validation of the effectiveness of the proposed denoising method.     

A hierarchical refinement algorithm for fully automatic gridding in spotted DNA microarray image processing

Gridding, the first step in spotted DNA microarray image processing, usually requires human intervention to achieve acceptable accuracy. We present a new algorithm for automatic gridding based on hierarchical refinement to improve the efficiency, robustness and reproducibility of microarray data analysis. This algorithm employs morphological reconstruction along with global and local rotation detection, non-parametric optimal thresholding and local fine-tuning without any human intervention. Using synthetic data and real microarray images of different sizes and with different degrees of rotation of subarrays, we demonstrate that this algorithm can detect and compensate for alignment and rotation problems to obtain reliable and robust results.     

一种基于前向纠错码的图像DNA加密存储算法

近年来DNA (DeoxyriboNucleic Acid) 存储发展迅速, 实现数字图像DNA存储和安全传输成为有待解决的问题。因此该文提出了一种面向DNA存储的基于前向纠错码的图像加密算法。首先使用动态约瑟夫遍历算法对图像像素点进行行置换和列置换, 以消除明文图像相邻像素之间的相关性。其次, 使用图像分解方法将明文图像分解为8个子图, 然后再重新组合, 实现了对图像像素值的置换, 从而进一步消除明文图像的纹理特征和破坏其统计学特征。再次, 对图像进行全局扩散, 使明文的微小变化以扩散的形式影响密文, 以抵抗差分攻击。最后使用可纠错DNA编码表将图像加密编码为DNA序列, 合成后进行存储。算法将明文图像加密成DNA序列并存储, 这种存储方式与传统存储介质相比更为安全。同时, 可纠错DNA码使得密文可以在DNA存储环境中可靠读取。该文使用3张常用图像包括lena_gray、peppers_gray、baboon_gray, 测试算法的安全性以及在DNA存储环境下的鲁棒性。仿真结果表明, 该方法可以有效抵御多种密码学攻击, 并且在DNA存储环境下对碱基错误和序列缺失等问题表现出良好的鲁棒性。