Typical remote sensing image processing requires massive data processing capabilities and storage facilities, which creates new challenges facing current remote sensing applications, especially for the multiple spectral images. CUDA (Compute Unified Device Architecture) programming framework is emerging as a promising platform for remote sensing image processing due to its common parallel computing support, flexibility, lower cost, and convenience with C programming language. In this paper we provide a prototypical implementation for remote sensing image processing on a CUDA support platform. This prototypical implementation includes the preprocessing and analysis of remote sensing image file format, calibration, CUDA programming and pixel computation and transferring, and results storing. Based on the implementation, a case study of soil moisture estimation is provided and it suggests the applicability and efficiency of CUDA computing platform in remote sensing image processing.
