Assessment of surface roughness based on super resolution reconstruction algorithm

In this work, an attempt has been made to use a super resolution image processing algorithm for preprocessing the images over and above existing image quality enhancement techniques. The improved quality images processed using a machine vision system have been used to assess the quality of the surfaces. To ensure the validity of the approach the roughness values quantified using these images are then compared with widely accepted standard mechanical stylus instrument values. Quantification of digital images for surface roughness is performed using two Fourier transform parameters (major peak frequency and principal component magnitude squared value) and the standard deviation of gray level intensity values. Then the group method of data handling (GMDH) technique was used to obtain an analytical relationship of the roughness parameters calculated using the digital surface image and the stylus instrument values. We present in this paper an analysis based on the comparison to make sure that the present approach of estimation of surface finish based on the digital processed image could be implemented in practice.     

The method of increasing COMPO contrast by linearization of backscattering characteristic η =f(Z)

The backscattered electron signal (BSE) in the scanning electron microscope (SEM) has been used for investigation of a specimen surface composition (COMPO mode). Creation of a material composition map is difficult because the dependence of backscattering coefficient η on the atomic number Z for Z > 40 is nonlinear. The method of increase in SEM resolution for the BSE signal by use of digital image processing has been proposed. This method is called the linearization of the η =f(Z) characteristic. The function approximating the experimental η =f (Z) dependence was determined by numerical methods. After characteristics linearization, the digital image in COMPO mode allows to distinguish between two elements with high atomic numbers if their atomic numbers differ by ΔZ = 1.     

Deconvolution improves colocalization analysis of multiple fluorochromes in 3D confocal data sets more than filtering techniques

Background and noise impair image quality by affecting resolution and obscuring image detail in the low intensity range. Because background levels in unprocessed confocal images are frequently at about 30% maximum intensity, colocalization analysis, a typical segmentation process, is limited to high intensity signal and prone to noise‐induced, false‐positive events. This makes suppression or removal of background crucial for this kind of image analysis. This paper examines the effects of median filtering and deconvolution, two image‐processing techniques enhancing the signal‐to‐noise ratio (SNR), on the results of colocalization analysis in confocal data sets of biological specimens. The data show that median filtering can improve the SNR by a factor of 2. The technique eliminates noise‐induced colocalization events successfully. However, because filtering recovers voxel values from the local neighbourhood false‐negative (‘dissipation’ of signal intensity below threshold value) as well as false‐positive (‘fusion’ of noise with low intensity signal resulting in above threshold intensities), results can be generated. In addition, filtering involves the convolution of an image with a kernel, a procedure that inherently impairs resolution. Image restoration by deconvolution avoids both of these disadvantages. Such routines calculate a model of the object considering various parameters that impair image formation and are able to suppress background down to very low levels (< 10% maximum intensity, resulting in a SNR improved by a factor 3 as compared to raw images). This makes additional objects in the low intensity but high frequency range available to analysis. In addition, removal of noise and distortions induced by the optical system results in improved resolution, which is of critical importance in cases involving objects of near resolution size. The technique is, however, sensitive to overestimation of the background level. In conclusion, colocalization analysis will be improved by deconvolution more than by filtering. This applies especially to specimens characterized by small object size and/or low intensities.     

A simple computer program to obtain digital pseudo-colored high resolution electron microscopy images

Computer programs which allow one to simulate high resolution electron microscopy images generally deliver the output on a line printer with the overprinting technique. Although this method suffers from poor resolution and limited contrast dynamics, it is still in widespread use. A more effective way is to represent the images in digital form on a color video display, and several solutions have been proposed recently, which often, however, make necessary use of rather expensive pictorial video terminals. In this work we describe a simple program which allows one to represent the output of a multislice program (projected potentials, lattice images, and diffraction patterns) on the display of an inexpensive color video terminal. Useful operations on the image, such as color selection over a wide range, image magnification, and filtering (to enhance specific details of interest) have been implemented. The displayed image can be either photographed from the screen or recorded on a good quality hard copier.     

Image restoration for TV‐scan moving images acquired through a semiconductor backscattered electron detector

A semiconductor backscattered electron (BSE) detector has become popular in scanning electron microscopy session. However, detectors of semiconductor type have a serious disadvantage on the frequency characteristics. As a result, fast scan (e.g. TV‐scan) BSE image should be blurred remarkably. It is the purpose of this study to restore this degradation by using digital image processing technology. In order to improve it practically, we have to settle several problems, such as noise, undesirable processing artifacts, and ease of use. Image processing techniques in an impromptu manner like a conventional mask processing are unhelpful for this study, because a complicated degradation of output signal affects severely the phase response as well as the amplitude response of our SEM system. Hence, based on the characteristics of an SEM signal obtained from the semiconductor BSE detector, a proper inverse filter in Fourier domain is designed successfully. Finally, the inverse filter is converted to a special convolution mask, which is skillfully designed, and applied for TV‐scan moving BSE images. The improved BSE image is very effective in the work for finding important objects. SCANNING 31: 229–235, 2009. © 2010 Wiley Periodicals, Inc.     

Influence of topography and specimen preparation on backscattered electron images of bone

Backscattered electron (BSE) images of bone exhibit graylevel contrast between adjacent lamellae. Mathematical models suggest that interlamellar contrast in BSE images is an artifact due to topographic irregularities. However, little experimental evidence has been published to support these models, and it is not clear whether submicron topographical features will alter BSE graylevels. The goal of this study was to determine the effects of topography on BSE image mean graylevels and graylevel histogram widths using conventional specimen preparation techniques. White-light interferometry and quantitative BSE imaging were used to investigate the relationship between the BSE signal and specimen roughness. Backscattered electron image graylevel histogram widths correlated highly with surface roughness in rough preparations of homogeneous materials. The relationship between BSE histogram width and surface roughness was specimen dependent. Specimen topography coincided with the lamellar patterns within the bone tissue. Diamond micromilling reduced average surface roughness when compared with manual polishing techniques but did not significantly affect BSE graylevel histogram width. The study suggests that topography is a confounding factor in quantitative BSE analysis of bone. However, there is little quantitative difference between low-to-moderate magnification BSE images of bone specimens prepared by conventional polishing or diamond micromilling.     

Strain mapping accuracy improvement using super‐resolution techniques

Super‐resolution (SR) software‐based techniques aim at generating a final image by combining several noisy frames with lower resolution from the same scene. A comparative study on high‐resolution high‐angle annular dark field images of InAs/GaAs QDs has been carried out in order to evaluate the performance of the SR technique. The obtained SR images present enhanced resolution and higher signal‐to‐noise (SNR) ratio and sharpness regarding the experimental images. In addition, SR is also applied in the field of strain analysis using digital image processing applications such as geometrical phase analysis and peak pairs analysis. The precision of the strain mappings can be improved when SR methodologies are applied to experimental images.     

Auto-correlation analysis of high resolution electron micrographs of near-amorphous thin films

In an attempt to characterize the degree of local ordering of amorphous, near-amorphous and microcrystalline materials from information contained in high resolution electron micrographs, a new method is considered in which auto-correlation functions are obtained from small image regions and then correlated in rotation. By finding the rotation which gives the best correlation with a particular local correlation function showing features of interest and then adding the local auto-correlation functions, an assessment of the local order for the whole image is achieved as a basis for consideration of the form and extent of the order in the specimen. The method has been applied to high resolution transmission electron microscope images recorded in digital form by use of a TV-based image intensifier and digital frame store and later analysed by use of an image processing system. The samples used were thin carbon films having different degrees of graphitization. Even the least-ordered carbon films showed significant local correlations.     

Evaluation of three methods for retrospective correction of vignetting on medical microscopy images utilizing two open source software tools

Correction of vignetting on images obtained by a digital camera mounted on a microscope is essential before applying image analysis. The aim of this study is to evaluate three methods for retrospective correction of vignetting on medical microscopy images and compare them with a prospective correction method. One digital image from four different tissues was used and a vignetting effect was applied on each of these images. The resulted vignetted image was replicated four times and in each replica a different method for vignetting correction was applied with fiji and gimp software tools. The highest peak signal-to-noise ratio from the comparison of each method to the original image was obtained from the prospective method in all tissues. The morphological filtering method provided the highest peak signal-to-noise ratio value amongst the retrospective methods. The prospective method is suggested as the method of choice for correction of vignetting and if it is not applicable, then the morphological filtering may be suggested as the retrospective alternative method.     

Deconvolution of scanning electron microscopy images

This paper describes a method of removing blurs in scanning electron microscopy (SEM) images caused by the existence of a finite beam size. Although the resolution of electron microscopy images has been dramatically improved by the use of high-brightness electron guns and low-aberration electron lenses, it is still limited by lens aberration and electron diffraction. Both are inevitable in practical electron optics. Therefore, a further reduction in resolution by improving SEM hardware seems difficult. In order to overcome this difficulty, computer deconvolution has been proposed for SEM images. In the present work, the SEM image is deconvoluted using the electron beam profile estimated from beam optics calculation. The results show that the resolution of the deconvoluted image is improved to one half of the resolution of the original SEM image.     

Spatial filtering of electron micrographs of negatively stained alpha-amylase crystals

Images of the alpha-amylase molecule from pig pancreas have been obtained in two projections by the application of spatial filtering techniques to electron micrographs of negatively stained microcrystals of the enzyme. The Fourier filtering was performed on a PPD 11/40 digital computer after microdensitometry on an Optronics P-1000 photoscan system. The set of procedures and programs we employed are described herein. We compared these images with equivalent images obtained by three-dimensional X-ray diffraction analysis using conventional isomorphous replacement at approximately twice the resolution. We find that there is quite good agreement between the two kinds of images, and that a number of gross structural features compare quite well. We conclude that the combination of electron microscopy with digital spatial filtering when applied to protein microcrystals yields very respectable results.     

Crystals with fivefold symmetry

A digital processing system has been applied to the signals of a multiple detector system for secondary (SE) and backscattered electrons (BSE) in a SEM. The system provides the usual contrast enhancement procedures, Fourier transform and correlation and, in addition, the summation, subtraction and division of images from different detectors. The difference signal of two SE detectors can be used to reconstruct the local surface tilt and the surface profile, and a subtraction of a BSE image from a SE image allows one to extract the pure surface information. Methods for correcting image shifts of sequentially recorded micrographs have been applied by making use of a Fourier transform or a cross-correlation.     

Methodology of improvement of radiometric quality of images acquired from low altitudes

Low-altitude photogrammetry studies have been more and more popular in the mapping of small areas (up to 10 thousand hectares). UAV flights can be consider as an attractive low-cost alternative solution for the photogrammetric studies. However, in this type of platforms images are frequently captured by digital compact cameras. Despite high resolution, the images taken with these cameras have a relatively low radiometric quality. While photogrammetric software for processing images obtained via sensors mounted on UAVs and the possible applications of the GPS RTK system for determining projection centres are constantly developing, the majority of studies nowadays still require digital aerial triangulation based on transferring and measuring tie points on subsequent images of the same surface fragment. A method for improving the quality of low-altitude image data is presented in this article. In order to improve the image radiometry, filtration in frequency domain was applied. This solution made it possible to enhance the reflection from objects in the images and at the same time reduce the impact of poor lighting on local contrast. The proposed method comprises two variants of radiometric correction, each of these depending on the quality of the pictures. The effectiveness of the method has been proven by adjusting three image blocks with different levels of radiometric quality before and after filtration, as well as a comparative analysis of the aerial triangulation results.     

Application of the Laplacian filter to high-resolution enhancement of SEM images

Certain digital image-processing methods, which are useful for nonperiodic structural images, have been applied to high-resolution SEM images for the improvement of resolution. Samples utilized in the present study consisted of magnetic tape coated with gold, T4 phage coated with gold-palladium, and uncoated specimens of Prolamellar body (PLB) in Cucurbita moschata. These images were blurred and otherwise disturbed by electronic noise, though the images were taken at the limit of efficiency of intrinsic instrument. The major image-processing tool was the Laplacian filter, which subtracts the Laplacian from the original image. Noise, which is a serious problem in digital processing of high-resolution SEM images, was suppressed by the nonlinear type smoothing method. Also, the noise was evaluated by an autocorrelation function and a power spectrum of the image. By using these methods of “deblurring” and noise removal, we achieved better resolution, and structural details of our biological specimens were revealed.     

A simple method for the acquisition of high-quality digital images from analog scanning electron microscopes

A method is described for converting video signals of analog scanning electron microscopes (SEMs) into digital images of high quality. A plug-in card commercially available for personal computers is used for the on-line analog/digital conversion. A Windows application program written by the authors, together with low-level software drivers supplied with the plug-in card, allow digital images to be recorded, to be displayed simultaneously on the computer monitor and to be saved as a file in a standardized format. Compared to conventional photographic images obtained from the SEM camera system, the digital images possess superior sharpness of outline, excellent image definition, diminished noise and well-defined grey-scale tones. This method provides SEM images of high quality for less than $1000 from most older analog SEMs. In addition, the advantages of digital image processing can be applied to analog SEMs, including contrast enhancement, digital filtering and multichannel recording.     

An evaluation of two-photon excitation versus confocal and digital deconvolution fluorescence microscopy imaging in Xenopus morphogenesis.

The ability to visualize cell motility occurring deep in the context of opaque tissues will allow many currently intractable issues in developmental biology and organogenesis to be addressed. In this study, we compare two-photon excitation with laser scanning confocal and conventional digital deconvolution fluorescence microscopy, using the same optical configuration, for their ability to resolve cell shape deep in Xenopus gastrula and neurula tissues. The two-photon microscope offers better depth penetration and less autofluorescence compared to confocal and conventional deconvolution imaging. Both two-photon excitation and confocal microscopy also provide improved rejection of "out-of-focus" noise and better lateral and axial resolution than conventional digital deconvolution microscopy. Deep Xenopus cells are best resolved by applying the digital deconvolution method on the two-photon images. We have also found that the two-photon has better depth penetration without any degradation in the image quality of interior sections compared to the other two techniques. Also, we have demonstrated that the quality of the image changes at different depths for various excitation powers.     

基于半盲解卷积复原的高分辨率视网膜成像系统

为获得高分辨率视网膜图像,建立了基于自适应光学的视网膜成像系统,并以成像时获得的残余像差作为图像复原的估计参数,通过半盲解卷积进行图像复原以获得高质量图像.通过Hartmann-Shark波前传感器和微机械薄膜变形镜组成的自适应光学系统对活体人眼像差进行测量与校正,并在成像时记录系统残余像差,据此重建光学传递函数作为图像复原模型初始参数估计,对获得的视网膜图像进行条件约束迭代半盲解卷积复原,消除像差对成像质量的影响,从而得到高分辨率视网膜图像.实验表明,系统获得的图像经该方法处理后可获得较满意视网膜图像,图像质量提高近一倍,成像成功率由38％提高至78％,成像时间缩短为原来的1/7.该方法在满足使用要求的前提下有效缩短了校正时间,提高了成像的成功率,提升了系统的适用范围.     

Theoretical explanation of the relationship between backscattered electron and x-ray linear attenuation coefficients in calcified tissues

X-ray absorption and backscattered electron (BSE) microscopies are two commonly used techniques for estimating mineral contents in calcified tissues. The resolution in BSE images is usually higher than in x-ray images, but due to the previous lack of good standards to quantify the grey levels in BSE images of bones and teeth, x-ray microtomog-raphy (XMT) images of the same specimens have been used for calibration. However, the physics of these two techniques is different: for a specimen with a given composition, the x-ray linear attenuation coefficient is proportional to density, but there is no such relation with the BSE coefficient. To understand the reason that this calibration appears to be valid, the behaviour of simulated bone samples was investigated. In this, the bone samples were modelled as having three phases: hydroxyapatite (Ca₁₀(PO₄)₆(OH)₂), protein, and void (either empty or completely filled with polymethylmethacrylate (PMMA), a resin which is usually used for embedding bones and teeth in microscopic studies). The x-ray linear attenuation coefficients (calculated using published data) and the BSE coefficients (calculated using Monte Carlo simulation) were compared for samples of various phase proportions. It was found that the BSE coefficient correlated only with the x-ray attenuation coefficient for samples with PMMA infiltration. This was attributed to the properties of PMMA (density and mean atomic number) being very similar to those of the protein; therefore, the sample behaves like a two-phase system which allows the establishment of a monotonic relation between density and BSE coefficient. With the newly developed standards (brominated and iodinated dimethacrylate esters) for BSE microscopy of bone, grey levels can be converted to absolute BSE coefficients by linear interpolation, from which equivalent densities can be determined.     

自适应窗口形状的中值滤波

中值滤波去除噪声的同时,难免出现含有线形图像的边缘轮廓信息跟随丢失的情况。为了达到滤除图像噪声最大化,有用信息损失最小化的目的,根据图像边缘特征选择或设计了适当大小和形状的窗口来进行中值滤波等操作。一方面,创新性地提出将Hough变换应用到车轮裂缝CT图像的滤波窗口形状的选择上,针对单一方向轮廓的图像,利用Hough变换检测出裂缝的方向,从而采用与裂缝形状相应的窗口对裂缝进行有针对性滤波,将该方法与传统方法进行了对比实验,并将60层滤波后切片图像堆栈得到了三维图像;另一方面,针对含有多个方向轮廓的图像,为了进一步改善滤波效果,提出根据像素梯度设计斜向滤波器,对多方向的线对进行滤波,数据显示经改进方法滤波后图像的峰值信噪比(PSNR)较传统的中值滤波提高了4~6,结构相似性(SSIM)提高了1%~2%左右,最后用683层荞麦切片CT图堆栈得到三维图像,对比滤波前后图像,可知该方法滤波效果良好。