An Automated Method for Container Counting in Satellite Images Based on Grid Analysis and Shadow Recognition

This study developed a novel and reusable technique aimed at automatically estimating the number of containers in port storage yards using satellite images, namely, an automated method for container counting in satellite images based on grid analysis and shadow recognition. Unlike traditional machine learning methods that rely on large-scale labeled datasets, this method does not require extensive data collections, significantly reducing the implementation threshold and costs. By dividing the container yards into grids, researchers were able to segment large-scale yard images into smaller blocks, allowing for individual block analysis. This not only improves processing speed but also enhances estimation accuracy. Converting images from the RGB color space to the HSV color space allows the algorithm to more accurately analyze and identify containers and their shadows. Furthermore, this study leverages the features of container shadows, which exhibit specific shapes and orientations in satellite images, to estimate container stacking heights. The number of containers in each grid is determined by comparing the shadow area with the known shadow area of a single container, making the overall calculation process more intuitive and reliable. Compared to the U-Net model based on semantic segmentation, this method significantly improves efficiency, reducing average program runtime by 72.90%, and greatly reducing the mean absolute error (MAE) and root mean square error (RMSE) of predictions. Compared to manual counting methods, the average time consumption of this method is also reduced by 69.75%, significantly enhancing efficiency. In summary, this integrated method of grid division, HSV color conversion, and shadow analysis provides a highly cost-effective and accessible tool for third-party researchers and operators. With appropriate satellite images, it enables real-time estimation of the number of containers in port storage yards.