ВИЯВЛЕННЯ МІН ЗА ДОПОМОГОЮ РОБОТОТЕХНІЧНИХ СИСТЕМ ІЗ ВИКОРИСТАННЯМ МУЛЬТИСПЕКТРАЛЬНИХ ВІДЕОЗОБРАЖЕНЬ

Relevance. Landmine detection using robotic systems with multispectral video imagery is driven by the critical need for safe, efficient, and highly accurate demining methods on post-conflict territories, where traditional approaches fail to provide sufficient speed and reliability in detecting explosive devices. Object of research: the process of remote search and detection of anti-personnel and anti-tank mines performed by robotic complexes utilizing computer vision in visible and thermal ranges. Purpose of the article. Development of a methodology for multispectral spatial analysis based on the synergy of spatial, structural (textural), and thermal characteristics of objects. Research results. The article details the physical nature of visibility differences (contrast) formation between explosive devices and their surroundings, principal configurations of sensor equipment, algorithmic stages of video stream processing in the onboard computer system within operational time frames, and the degree of external conditions' impact on identification effectiveness. It is established that the combined use of RGB cameras and thermal imagers (RGB-IR fusion) significantly improves the proportion of correctly detected targets under adverse observation conditions. Conclusions. The presented data demonstrate substantial potential for applying unmanned ground and aerial vehicles for multispectral monitoring in humanitarian demining operations. Scope of application of the obtained results: mobile robotic systems for multispectral detection of explosive objects.