APPLICATION OF ARTIFICIAL INTELLIGENCE ALGORITHMS OPENCV AND LOW-CEILING PHOTOGRAMMETRY TO ASSESS THE QUALITY OF A SAND DEPOSIT AND PREPARATION OF THE CORRELATION MODEL

• Published in: International Multidisciplinary Scientific GeoConference SGEM Vol. 19; no. 2.2; pp. 31 - 38
• Main Authors: Kardasz, Ewa, Sołtysiak, Marek, Jarzębski, Maciej, Iljaszewicz, Paweł, Kardasz, Piotr
• Format: Conference Proceeding
• Language: English
• Published: Sofia Surveying Geology & Mining Ecology Management (SGEM) 01.01.2019
• Subjects: Aerial surveys; Algorithms; Artificial intelligence; Cameras; Computer programs; Correlation; Electromagnetic radiation; Fields; Frequency ranges; Genetic algorithms; Geological surveys; Geologists; Geology; Geomorphology; Groundwater; Hydrologists; Hydrology; Image acquisition; Measurement; Measurement techniques; Measuring instruments; Methods; Mining; Morphology; Neural networks; Normalized difference vegetative index; Open pit mining; Pattern recognition; Photogrammetry; Properties; Quality assessment; Quotients; Radiation; Reflectance; Reflection; Remote sensing; Sand; Software; Spectra; Surveying; Unmanned aerial vehicles; Vegetation; Vegetation index; Wave reflection; Wavelength; Wavelet transforms; Poland
• ISSN: 1314-2704
• DOI: 10.5593/sgem2019/2.2/S09.005

In recent years, there has been observed an intensive development of consumer unmanned aerial vehicles (UAV). Their advantage is a relatively low cost compared to typical aircraft constructions. The UAV equipment in recording devices allows them to be used for surveying, volumetric measurements, vegetation index calculations (including the normalized difference vegetation index) and thermovision analyzes. Due to the high resolution, the materials obtained from UAV can be a source of information, among others for designers, hydrologists, geomorphologists, geologists and others. The acquired images may also be subject to advanced analyzes using, among others, methods of artificial intelligence. The article presents the results of the analysis of photos acquired during a mission over the exploited sand pit. The aim of the research was to check if low-ceiling photogrammetry would be useful to assess the quality of a sand deposit. The results of geological surveys were taken into account in addition to the photos obtained from the UAV. FLIR Tools software was used for analysis, as well as specialized algorithms based on OpenCV software. We plan that the final goal will be to develop a methodology for performing non-invasive measurements to enable monitoring of the deposit. The method used allows remote sensing of the electromagnetic radiation interaction with the examined matter. Part of the radiation is selectively absorbed, reflected or transmitted. A bandwidth compatible with the characteristics of our camera was used for remote sensing. We assume the use of other cameras to measure other frequency ranges. The basic measure is the spectral reflection coefficient, which defines the quotient of energy reflected to the electromagnetic spectrum falling in the given range. The function of the spectral reflection coefficient from the wavelength is the spectral reflection curve, also referred to as spectral characteristic. The spectral properties of the tested deposit depend on the structure, morphology and properties of the deposit. The collected information was analyzed using several available methods by selecting the optimal genetic algorithm and neural networks to analyze the obtained data. The aim of the measurements is guided by the idea of proper interpretation of the obtained results and implementation of them in the form of ready-made patterns related to the researched fields. In comparison to the results obtained in a conventional way, the presented method is less expensive and allows for commercial implementation. The aim of the research is to build a model that would define the dependence of the factors studied. Then, based on the constructed model, it will allow to obtain a correlation between the occurring values and the measured parameters. The model will allow for the commercial use of research and reduction of the implementation costs of measurements. The spectral detection method will in many cases replace the previous measurement techniques, providing a lot of valuable information in a very short time. Measurements are easy to make and unstructured, and the available measuring instrument is relatively cheap. The mentioned advantages of the measurement and indirect evaluation of the status of the studied fields are the reasons to choose this method of measurement as a secondary study tool. An important aspect, in addition to the measurement itself, is the development of indicators and their interpretation based on the data obtained. It will be possible to use the method in other geological application areas. Here, the users of the method will be the owners of opencast mining plants and the bodies of geological and mining administration.