Автоматизована побудова цифрової моделі мікроповерхні об’єкта за стереопарою цифрових РЕМ-зображень

Abstract

The purpose of this work is to develop and study the method of automated construction of a digital model of the micro relief of an object's surface using a stereo pair of digital SEM images taking into account the specifics of a SEM survey and studying the accuracy of digital modeling of a micro surface of an object, depending on the type of input point generation. Methodology. Research is based on the processing of three types of data models. The test model is derived from the measurement of digital images of SEM stereo pairs. Other models are constructed from the points generated on the left image of the SEM stereo pairs by a regular grid and by a gradient method. The identification of the corresponding points of the stereo pair is performed by a correlation method in stages with the use of search windows of various sizes. To obtain the spatial coordinates of the micro relief points of the object's surface with the accuracy required for the researchers, the values of the geometric distortions inherent in the SEM image are taken into account. In order to eliminate some abnormal values of the heights of a 3D model, an adaptive median filtering procedure is used. The estimation of the accuracy of the values of the heights of the points of the micro relief of the surface of the object is performed by comparison with the test model. Results. The possibility of a stereo pair of digital SEM images in the automated mode is established to reproduce the micro relief of an object's surface with the requirements of both the accuracy of determining the spatial coordinates of the points and the structure of the micro surface of the object. The developed algorithm is implemented in MATLAB. Due to the dense set of points, both models allow to correctly model the small elements of the structure of the micro surface. The number of points in which the difference between the values of the heights of the test and the studied models are within the tolerance 5Z ~ ± 2 mu is 88–89 %. The estimation of the algorithm's time for identifying the corresponding points of the stereo pair using the 2–3-search windows of various sizes is performed. Scientific novelty. The authors developed a method of step-by-step correlation identification using search windows of various sizes is accompanied at all stages by the author's software. Evaluation of the effectiveness of this technique has shown the feasibility of its practical application. Practical significance. The application of this technique allows to automate the process of constructing a digital model of micro relief of the object's surface based on the SEM stereo pairs of digital images. The results of the experiment confirm the effectiveness of the proposed algorithm and a significant reduction in the construction time of the digital model of the micro relief compared with the time measurement of points coordinates by the operator in manual mode.