11. Study on methods for processing large remote sensing datasets using the high-performance image processing system Pixel Factory

Abstract

In recent years, remote sensing data in Vietnam has been increasingly applied, especially high-resolution and very high-resolution remote sensing data at regional and national scales. At these scales, data volumes are becoming larger, requiring rapid processing times. During project implementation, the fast and accurate processing of large volumes of orthorectified remote sensing imagery over wide areas remains a challenging and urgent issue to address. The high-performance remote sensing data processing system (Pixel Factory) plays a crucial role in producing digital orthophotos thanks to its advantages in batch image data processing, ensuring high accuracy and efficiency. This paper is based on imagery acquired from the SPOT 6 satellite, applying the Pixel Factory (PF) system to process high-resolution remote sensing data. It describes in detail the automated data processing workflow, from input data preparation to final product generation. At the same time, the paper analyzes existing issues and proposes methods to improve the efficiency of processing large remote sensing datasets. The results show that using the PF system requires only 5 - 7 Ground Control Points (GCPs) per block/scene; In some difficult cases of measurement or selection on the imagery, only 3 - 4 GCPs may be sufficient. Meanwhile, according to current technical standards as well as when using conventional commercial image processing software, the minimum number of GCPs required is 10 - 12 per scene. This demonstrates that the PF system significantly reduces the number of GCPs required and shortens processing time when handling large blocks of remote sensing imagery.