Abstract
In that work a describing of automated system is made an introduction, which allow to identify particles, that are look after on electron-microscopic images, and to determine their size, area and length of borders independently. Particle identification is based on a threshold criterion, which has low computational complexity, but has proven to be quite effective when applied to SEM and TEM images. Visual control of correct particle identification is facilitated by colorizing images. The proposed software system has a high performance, and the processing of a typical SEM imagine usually take less than 10 seconds. The proposed approach was tested on different vacuum condensates and shown high efficiency for different SEM and TEM images. It is effective for simple pictures, which are characteristic for samples, formed by melting of sufficiently thick polycrystalline films. This approach can also be used for quantitative processing of images obtained from films condensed by the vapor-liquid mechanism. Usually such images are difficult for automatic processing. Data on the size dependence of the most probable particle radius formed during melting tin films, full width at half maximum of histograms of their size distribution and excess energy, stimulating the process of de-wetting of films were received by using the proposed software. It is shown that particle size distribution for samples obtained by melting initially continuous polycrystalline films is single-mode and has normal character. The ratio of the full width at half the height of the distribution histogram to the most probable particle radius does not depend on the film thickness. Applying offered method for alloys films let to establish the effect of composition on films’ de-wetting and to show, that sequential vacuum condensation and subsequent melting are a easy method of forming arrays both single-component particles and alloys particles.
Publisher
V. N. Karazin Kharkiv National University