Small Target Detection and Geolocation Software in Satellite-Based Remote Sensing Image

Due to the limited computing power and storage capacity of satellite equipment, current pixel level small target detection methods cannot operate effectively. This paper proposes a small target detection and geolocation software. To solve the difficulty of detecting pixel-level small targets based on shape, we use K-means clustering based on feature colors for detection. To determine the latitude and longitude of specific locations in remote sensing images captured by satellites, we use maps for image registration to calculate the pixel coordinates of small targets. Finally, the folium library is used to parse the latitude and longitude of that location. The experimental results are consistent with manual annotation and show high accuracy.