At two dimensional digital analisis with many points (Mapping), the following conditions and accuracies of sampling data are discussed; 1. Relation between the sampling time for one picture element and the accuracy. 2. Conditions for the recognition of a image contrast. 3. Conditions for the recognition of a edge and boundary. 4. Relation between the numbers of picture elements and the accuracy. 5. Phenomenon of the transformation of a image shape in the case of probe size which is not so small as compared with the specimen structure size.
A fundamental set of image processing methods are presented. They are selected from conventional methods in view of fundamental pattern recognition, not image enhancement, image transmission nor storage of images.
The fundamental study of digital image processing, in relation to the scanning electron microscopy seems to have provisionally been completed. Since the SEM image is essentially of electric signal, the on-line digital image processing system linked to the SEM should find unique applications different from the simple digital image processing. This paper describes several useful techniques applied. to the practical on-line processing. Techniques of interpolation, signal averaging, Laplacian filter, smoothing by averaging along edges, homomorphic filter, etc. have been used to reduce the unfavorable effects in SEM images. From our experimental results, we have confirmed a potential ability of on-line digital image processing system, which has not yet been established.
Two kinds of image processing techniques can be applied to the high resolution electron microscope (HREM) images : one in the real space (averaging of periodic images) and the other in the Fourier space. There are several methods in the latter, such as ring masking, window masking and a CTF (contrast transfer function) compensation using the Wiener filter. The Wiener filter is extremely useful for the HREM image processing, because the filter can improve the CTF, and thus provide improve images. The image contrast can partially be reversed and that in the high frequency region can be enhanced by the Wiener filter. A practical method for the Wiener filter is explained by using a HREM image of multiply twinned Ag particle. An application is shown with a high temperature superconducting material, and the other techniques are also demonstrated with various images.
An technique of computer image processing has been applied to the X-ray images of electron probe micro-analyser. Using the averaging method to reduce image noise, the quality of X-ray images was improved and thus the segregation of Ti along the boundary region between ceramics and maraging steel was revealed. A new technique of non-destructive 3-dimentional quantitative analysis of surface layer has also been studied and applied to the observation of the distribution of oxide particles of TiO2 in metal titanium. In this method, a series of X-ray images taken by increasing the acceleration voltage to increase the depth of X-ray generation in the specimen provided the 3-dimentional distribution of chemical elements.
In the analysis of high-quality steels, requirements have shifted to the analysis of elemental distribution and compound structure and size. To serve this purpose, Nippon Steel developed a method for analyzing the elemental distribution in microscopic and macroscopic regions and a computer-aided X-ray microanalyzer (CMA), which may be called a comprehensive state analyzer that combines the basic principle of the electron probe X-ray microanalyzer (EPMA) with the data processing capability of computers. By using CMA two-dimensionally collected characteristic X-rays and electron signals generated at many points, on the sample surfaces are image-reconstructed by the computer to characterize the material under analysis. In addition to iron and steel, the CMA has been successfully applied to a variety of new advanced materials, such as nonferrous metals, ceramics, carbon materials and electronic materials.
We have recently developed a high speed mapping system (Hsystem) and a scanning image processing system (SIPS). The Hsystem comprises a digital scan converter, a color controller, and a color CRT. The digital scan converter collects various signals such as secondary electrons, backscattering electrons, X-ray, and others by means of either beam scanning or stage scanning at a very high speed. The image information is displayed on the color CRT after realtime processing by the color controller. The SIPS is a computer system that dynamically controls EPMA by means of a 32-bit super minicomputer. The system has an automatic analytical software including collecting mapping data from the Hsystem or directly from the EPMA, and permits highly sophisticated image analysis of mapping data as the image and the own data of EPMA.
Scanning Auger Microscopy is an established surface analysis technique in which specimens are excited with a focused primary electron beam and emitted Auger electrons are detected with an energy analyzer. SAM provides the topographical and elemental information of specimen surfaces also to in a microscopical scale. Image processing softwares enhance the perception and understanding of the data in micro analysis. The accuracy of the Auger data is related to the signal-to-noise ratio and the detection limit.
The application of digital image processing to SIMS data has not yet to realized its full potential. Quantitative mapping of the element distribution on the sub-micron scale with high sensitivity is performed only by using SIMS. However, in order to utilize SIMS practically, many problems such as ion matrix effects, topographical effect, nonuniform sputter removal of matrix layers and implanted primary ions which might cause chemical effects, should be solved. This report is to discuss the problems involved in image acquisition, correction, analysis and display, and show how digital image processing techniques can be used extend the capabilities of SIMS in the analysis of materials which are heterogeneous on the sub-micron scale.
Rapid data acquisition is required for the two dimentional graphic data analysis by using scanning sensors or electrodes in the X-Y direction. In order to satisfy this requirement, a high speed X-Y scanning system has been developed, in which data can be sampled every 40μs and 3 million data can be graphically analyzed in 10 minutes. The localized corrosion behavior was studied by using a microelectrode. The experimental results showed that this scanning microelectrode system can be used to detect the localized corrosion on alloys and also fine defects of coated materials. It was also found that the newly scratches formed can be detected by using the exo-electron detector as a sensor.
Recently the observation technique using SEM backscattered electron images becomes important in analyzing core-loss mechanisms of silicon steels. Particularly the high voltage SEM is expected as auseful mean for observing glass-coated products. We have developed a new technique called “line-sampling stroboscopy” which enables the dynamic SEM observation for the periodic phenomena driven at line frequencies. Combining these techniques, the dynamic magnetization process in silicon steel products can be observed in practical drive conditions.
Natural soils on collars have been used to evaluate detergency and this is known as the most reliable method. However, quantitative measurements has not been made because of randomness of natural soils. Recently we developed a new quantitative detergency evaluation method using an image analyser to replace the visual judgement. There is a linear relationship between the actual amount of soils and the degree determined by our equation based on Kubelka-Munks'. This method has a good correlation with the traditional visual judgement.