Software for optical recognition of micro- and nano-objects in solids and colloidal solutions
Autor: | L. Diachenko, P. M. Fochuk, Sergiy Ostapov, O. Val, E. Minov, Yu. B. Khalavka, O. Kopach |
---|---|
Rok vydání: | 2015 |
Předmět: |
Microscope
Materials science business.industry ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Cognitive neuroscience of visual object recognition law.invention Set (abstract data type) Crystal Wavelet Software Planar law Computer Science::Computer Vision and Pattern Recognition Point (geometry) Computer vision Artificial intelligence business |
Zdroj: | SPIE Proceedings. |
ISSN: | 0277-786X |
DOI: | 10.1117/12.2228627 |
Popis: | This paper deals with the development of algorithms and software for optical recognition of growing defects in the semiconductor crystals and metal nanoparticles in colloidal solutions. Input information is a set of photographs from a microscope, as well as a short video-file with nanoparticle's tracks. We used the wavelet technology to filtering and image transformations. As a result of recognition the 3D image is formed with the point, linear and planar growing defects. Defects are sorted by size; different statistical characteristics are computed such as the defect’s distribution in layers and in the whole crystal. The system supports arbitrary rotations of the "crystal"; “cutting” by different planes and so on. The software allows you to track the movement of nanoparticles in colloidal solutions; to determine the local temperature and density of the solution. We proposed a new method for quantitative estimation of recognition quality. This method based on the "virtual crystal” model, which has predetermined parameters of the defect subsystem. The software generates a set of photographs, which used as the input information of recognition system. Comparing the statistical parameters of the input data with the recognition results, we can estimate the quality of recognition systems from different manufacturers. |
Databáze: | OpenAIRE |
Externí odkaz: |